Synchro-Phasor Data Conditioning and

Validation Project
Phase 1, Task 2 Report

Best Practice Recommendations for

Synchrophasor Systems: Administration, Planning
and Implementation, and Operation and Maintenance

Prepared for the

Office of Electricity Delivery and Energy Reliability,
Transmission Reliability Program of the U.S. Department of Energy
Under Contract No.DE-AC03-76SF00098

Prepared by:

Ken Martin, PI
Prashant Palayam
Iknoor Singh
Chen Sun
Xuanyu Wang
Project Manager: Dejan J. Sobajic

August 14, 2013

Acknowledgement

The work described in this report was coordinated by the Consortium for Electric Reliability
Technology Solutions, and funded by the Office of Electricity Delivery and Energy Reliability,
Transmission Reliability Program of the U.S. Department of Energy through a contract with
Electric Power Group, LLC administered by the Lawrence Berkeley National Laboratory.

Disclaimer

This document was prepared as an account of work sponsored by the United States Government.
While this document is believed to contain correct information, neither the United States
Government nor any agency thereof, nor The Regents of the University of California, nor any of
their employees, makes any warranty, express or implied, or assumes any legal responsibility for
the accuracy, completeness, or usefulness of any information, apparatus, product, or process
disclosed, or represents that its use would not infringe privately owned rights. Reference herein to
any specific commercial product, process, or service by its trade name, trademark, manufacturer,
or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or
favoring by the United States Government or any agency thereof, or The Regents of the University
of California. The views and opinions of authors expressed herein do not necessarily state or
reflect those of the United States Government or any agency thereof, or The Regents of the
University of California.
Ernest Orlando Lawrence Berkeley National Laboratory is an equal opportunity employer.

i
 TABLE OF CONTENTS

Acknowledgement ................................................................................................................................... i
EXECUTIVE SUMMARY............................................................................................................................. 1
CHAPTER 1. INTRODUCTION .................................................................................................................... 4
CHAPTER 2. ADMINISTRATION OF PHASOR MEASUREMENT SYSTEMS .................................................... 5
2.1 Project development administration ............................................................................................ 5
2.2 Synchrophasor system administration.......................................................................................... 5
2.3 Synchrophasor system documentation......................................................................................... 7
CHAPTER 3. PLANNING, DEVELOPMENT, AND IMPLEMENTATION ........................................................... 8
3.1 System planning............................................................................................................................. 8
3.2 System Development - Equipment Selection................................................................................ 11
3.2.1 Phasor Measurement Unit (PMU) selection........................................................................... 11
3.2.2 Communication equipment selection .................................................................................... 14
3.2.3 Timing equipment selection .................................................................................................. 14
3.2.4 Control center equipment selection ...................................................................................... 14
3.3 System Implementation ............................................................................................................... 15
3.3.1 Substation equipment installation......................................................................................... 15
3.3.2 Control center equipment installation................................................................................... 16
3.3.3 Measurement validation ....................................................................................................... 16
CHAPTER 4. OPERATION & MAINTENANCE OF PHASOR SYSTEMS .......................................................... 18
4.1 Operation and analysis................................................................................................................. 18
4.1.1 On-line data validation .......................................................................................................... 18
4.1.2 Off-line data validation.......................................................................................................... 19
4.1.3 Data analysis in operation ..................................................................................................... 21
4.2 System maintenance.................................................................................................................... 21
4.2.1 Hardware maintenance......................................................................................................... 21
4.2.2 Performance maintenance .................................................................................................... 22
4.2.3 Configuration management................................................................................................... 22
APPENDIX A. PMU AND SYSTEM INSTALLATION AND VALIDATION PROCEDURES.................................A-1
A.1 Introduction ...............................................................................................................................A-1
A.2 PMU compliance with C37.118.1 and calibration ........................................................................A-1
A.3 Substation installation requirements ..........................................................................................A-2

ii
 A.3.1 Check that the clock used for synchronization to UTC is on time and locked on time. Check
that the PMU correctly indicates when time is locked and that the lock is steady. ........................A-2
A.3.2 Confirm that the phasor measurement magnitudes are within 1% of input levels. Also
confirm voltages are within 1% + 1kV and currents within 1% + 100A of comparable substation
measurements..............................................................................................................................A-3
A.3.3 Confirm that the phasor measurement angle differences are within 1-3 degrees of
corresponding input signal angle differences. Also confirm the angles with comparable
measurements in the substation...................................................................................................A-4
A.3.4. Confirm that analog measurements are within 5% of other measuring devices in the
substation.....................................................................................................................................A-5
A.3.5. Confirm that digital status measurements report the correct Boolean state........................A-5
A.3.6 Substation installation confirmation documentation............................................................A-6
A.4 Control center installation requirements ....................................................................................A-6
A.4.1 Confirm that received data match the setup and the time stamps are within 3-seconds of the
local time......................................................................................................................................A-6
A.4.2 Catalog communication errors and show that the overall data loss is less than 0.1% over a 24-
hour period...................................................................................................................................A-7
A.4.3 Compare received signals with SCADA measurements .........................................................A-8
A.4.4 Validate data status indications .........................................................................................A-10
A.4.5 Compare received signals with EMS state estimation results..............................................A-10
A.4.6 Control center installation confirmation documentation....................................................A-11

APPENDIX B: TROUBLE RESOLUTION FOR INSTALLATION AND MAINTENANCE......................................B-1

B.1 Introduction ...............................................................................................................................B-1
B.2 Installation problem solving........................................................................................................B-1
B.2.1 PMU installation ..................................................................................................................B-1
B.3 OPERATION PROBLEM SOLVING .....................................................................................................B-4
B.3.1 Overview of synchrophasor data using C37.118 transmission ..................................................B-4
B.3.2 Operational problem troubleshooting......................................................................................B-6
B.3.2.1 Phase Loss.........................................................................................................................B-6
B.3.2.2 Resolution & Accuracy ......................................................................................................B-7
B.3.2.3 Input Scaling .....................................................................................................................B-9
B.3.2.4 Output Scaling...................................................................................................................B-9
B.3.2.4 Sync Loss at a PMU .........................................................................................................B-10
B.3.2.5 Intermittent Sync ............................................................................................................B-10
B.3.2.6 Derived Signals................................................................................................................B-11
B.3.2.7 Delays in signals received at the Control Center ..............................................................B-11
B.3.2.8 Data loss in signals received at the Control Center ..........................................................B-11
B.3.2.9 Data errors in signals received at the Control Center.......................................................B-12

Page iii
 Synchro-Phasor Data Conditioning and Validation Project

Phase 1, Task 2 Report

Best Practice Recommendations for Synchrophasor Systems:
Administration, Planning and Implementation,
and Operation and Maintenance

EXECUTIVE SUMMARY

This report presents results of Task 2 of the Data Validation and Conditioning project. The report covers
best practices for administration, planning and implementation, and operation and maintenance of
synchrophasor measurement systems. The best practice recommendations are based on the survey
carried out in Task 1 of this project and the experience of Electric Power Group, LLC (“EPG”) with
synchrophasor systems. The study starts with project administration and carries on through
maintenance.

Phasor systems have four essential components:

1. Measurement devices in substations.

2. Communications to bring the measurements from the substation to the Transmission Owner
(TO) control center.
3. Data processing in the control center.
4. Accumulation of data in archives or at a higher level such as an Independent System Operator
(ISO) control center.

These components are tightly coupled: a failure in one part will cause a failure in the end product and it
may be difficult to identify the root cause in a failure situation. It is important to start the phasor system
with a plan that covers all aspects. Planning should begin with the applications to assess their
requirements. The locations, measurements, and devices should be selected to support these
requirements. Consideration of communication, control center, data storage, and application
availability need to be factored in and may require some modifications to the original plan.

The best practice recommendations for administration, planning and implementation, and operation
and maintenance are summarized below.

Administration starts with assembling a design team and carries through with assuring timely
maintenance and support for upgrades. For the initial stages, it is important to bring together a team
that can deal with all aspects from measurements in substations to communications and finally to
computer-based applications. Phasor systems generate a lot of data that is sent and processed in real-
time, which creates more interdependencies than most utility systems in service today. It is important

Page 1
to deal with these complexities in the early stages. Coordination must continue through the
implementation stage to be sure the measurements are correctly identified, characterized, and
calibrated. As with any system, problems will occur and updates will be required. Ongoing
management needs to coordinate the activities of the responsible groups to maintain the system in
good operating order. A good documentation chain is needed to assure calibration is up to date and
restoration can be done effectively when there are problems. Best practices for administration include:

1. Multidisciplinary design team covering equipment selection, applications use, power systems,
information technology (IT), communications, and operations.
2. Design documentation and change management process.
3. Configuration management – connectivity of new equipment.
4. Performance management – data validity, equipment malfunction detection and restoration.

Planning should start with the applications to be sure the system meets their data requirements. This
includes measurement points (PMU locations) and equipment selection. During implementation it is
important to validate the measurements, preferably both with basic substation measurements and with
comparable measurements reported to the control center such as Supervisory Control and Data
Acquisition (SCADA). This step assures signals are correctly identified and calibrations are correctly
applied. Data processing in applications and archiving needs to be closely monitored initially to be sure
there are no systemic problems causing hidden failures. Best practices for planning and implementation
include:

1. Planning – phasor measurement unit (PMU) location, data rates, signal selection, application
requirements.
2. System Development – equipment selection (PMUs, timing equipment, etc.), communications,
and control center applications.
3. Implementation and installation – substation, control center.
4. Measurement validation – comparison with state estimator and SCADA, latency.

Once in operation, tools are needed to provide constant monitoring. Generally these are provided by
application vendors, but in some cases utilities develop their own applications to meet their special
needs. In both cases, a process for human monitoring is required to act on problems and note when
unusual situations occur. A regular maintenance program is advisable and can usually be established
along the lines of similar equipment under standard utility programs. Of special note are the high use of
data storage space and bandwidth by phasor systems; these particularly need to be monitored. Best
practices for operation and maintenance need to address:

1. Operations and Analysis of System Performance – function involves IT (network,

communications, data archiving) and power systems metrics.
2. On-Line Data Validation and Monitoring Tools – vendor provided as well as in-house
development to integrate with internal processes.
3. Off-line Data Validation and monitoring tools.

Page 2
 4. System Maintenance.
5. Configuration Management.
6. Data analysis – on-going monitoring of events to integrate with planning and operations,
including establishing alarm thresholds and operating procedures.

Each of these areas is discussed in the report. In addition, there are two appendices which provide
detailed guidelines.

 Appendix A provides procedures for validating the installation and measurements. It starts with
procedures for the PMU in the substation and continues on through the TO and ISO control
centers. It suggests alternative methods depending on what tools the utility has available. It
gives a range of accuracies to be expected.
 Appendix B is a troubleshooting guide. It details some background information on data quality
indications. It describes some common problems and details the analysis and solution. It both
provides a reference for the problems and methods for solving similar problems.

Page 3
CHAPTER 1. INTRODUCTION

This report presents results of Task 2 of the Data Validation and Conditioning project. This report covers
best practices for administration, planning and implementation, and operation and maintenance of
synchrophasor measurement systems. These best practice recommendations are based on the survey
carried out in Task 1 of this project and EPG experience with synchrophasor systems. The Task 1 report
detailed the results of a survey of synchrophasor projects at 20 companies in North America. That
survey found varying degrees of completion of projects and success with the results. This follow-on
report is intended to document the most successful approaches to these projects by way of best
practice recommendations. Most of the survey participants have been focused on infrastructure
deployment and hence their responses primarily addressed installation, measurement validation, and
problem resolution.

The overall project focus is data validation, making sure the data received from phasor systems is valid
and accurately represents the measured quantity so that the data can be relied upon in real-time
operations. Good project administration and system design are critical to having good and validated
data for use in operations. Systems are the product of everything that goes into them. If they are
poorly designed, problems will crop up more frequently and be more difficult to resolve. If the
governing administration is not effective, identifying problems, addressing them effectively and
resolution may be more difficult. Poor documentation can contribute to errors in use and repeated
problems. The report covers infrastructure as well as data from the standpoint of project
administration, planning and implementation, and operations of synchrophasor systems.

Page 4
CHAPTER 2. ADMINISTRATION OF PHASOR MEASUREMENT SYSTEMS

This section describes the administration of synchrophasor measurement systems and in particular
issues that make it different from administration of other systems. There are 2 principal phases of such
administration: first is the design and implementation and second is operation and maintenance. This
division is typical of most utility projects, so this is not unique. This chapter is organized with a section
on each of these phases. A section on documentation completes the administrative aspects.

2.1 Project development administration

Synchrophasor projects differ from many common utility projects in their high interdependency of
components ranging from the remote substation sites to the higher level control or monitoring centers.
Some relay projects involve direct high-speed connections but usually between substations. SCADA
involves reporting from substations to control centers, but at much slower rates. Special protection
schemes (RAS) often serves the wide-area but with intermittent signals. Synchrophasors are thus similar
to some existing systems, but require continuous, real-time communications at higher rates with active
signals over the entire grid. In addition, synchrophasor systems cover a wide-area involving multiple
companies that requires effective coordination and communications.

To be sure the scope of the project is fully considered during the project development phase, it is
recommended that a steering team is established for project administration. The team size will depend
on the size of project and the number of separate departments involved. For the transmission operator
(TO) which builds and operates the substations, the steering team should include representatives from
system planning, substation engineering (usually protection group), telecommunications (substations to
control center), IT/application support, and operations. For the system operator (ISO, RTO, etc.), the
team needs to include communications, IT administration (network, data storage), application support,
and operations. Other departments may need to be included if there are some special aspects of the
projects, such as added control functions. The makeup of the team is intended to make sure that all
aspects of the project are fully considered from the start. It is suggested that the team has co-leaders or
the team lead has rotation so that more than one person is familiar with the overall aspects of the
project. This helps assure continuity as staff changes, which can be important if the project carries on
for a longer period of time than anticipated.

Regular steering team meetings are recommended to deal with problems that come up as well as
sharing the progress and issues that have been resolved. This helps uncover issues that may become
problematic in the future, as well as keeping the project on track. Meeting records should be kept for
tracking problem notifications and resolutions.

2.2 Synchrophasor system administration

Usually each department manages their own equipment including changes in equipment and settings as
dictated by failures, upgrades, other installations, etc. This can lead to some difficulty since the
measurements can be considerably affected by changes in setting parameters even though the
measurements may look the same. In addition the operation of the system is highly dependent on

Page 5
many components in a chain, from the potential transformer/current transformer (PT/CT) through the
PMU, communications, routing, phasor data concentrator (PDC), storage, applications, and so on. When
there is a data problem at the user end, it could originate anywhere in the measurement chain making it
difficult to determine root cause and location of the problem. Synchrophasor systems need to be
managed as a SYSTEM and not individual devices, such as relays.

For these reasons, formation of an ongoing coordination team is recommended. This team should
include representatives from the following departments:

1. System planning (users).

2. IT – network and data storage (support).
3. Applications (support, configuration management, data analysis).
4. Operations (user).
5. Maintenance (support).

This team should have a clear charter that defines the individual responsibilities and the procedures to
be followed by the team members. The project team should have routine meetings for coordination
and for sharing issues and problems between different departments, at least in the initial stages. Once
the system is operating to its expected performance level, meetings could be reduced to an annual
review and special sessions to resolve issues. Duties of the coordination team include:

1. Examination and resolution of problems.

2. Coordination of changes in equipment and settings, both internal & with other organizations.
3. Determination of need for upgrades including added measurements, locations, and application
deployment.
4. Data storage and retention policy.
5. Recommendations on data sharing requests (decision rests on utility administration).
6. System security application (such as NERC Critical Infrastructure Protection {CIP}).
7. Documentation of activities & overall system documentation review.

Data users will usually be the first to discover problems or see the need for some changes. Typical users
include operators for real-time data and planning and engineering personnel for recorded data. In any
case, there should be a procedure to contact an appropriate representative of the coordination team for
further action. Standard company procedures for repair, maintenance, and project requests should be
sufficient for most cases. However for some issues, the coordination team can pull together a cross
discipline team that can solve more complex issues more quickly. In addition, often, problems extend
beyond company boundaries and coordination and communications among owners of all components
of the synchrophasor network is critical.

Participation on a coordination team will be an added responsibility to the usual duties for most staff.
Over time, there will be changes in personnel. Therefore rotation of the team lead is recommended to
build a broader knowledge base among the team members and assure continuity. Avoiding dependency

Page 6
on a single person, who may be on leave or may move to another job, is important for good functioning
of the whole project.

2.3 Synchrophasor system documentation

Documentation for synchrophasor systems should follow normal company procedures and standards.
Supplementary documents are recommended to help understand the following:

1. Overall system diagrams of phasor system.

2. Phasor system data flow & processing details.
3. Documentation of equipment settings (especially PMU).
4. Primer on phasor estimates from a PMU.

The overall system and data flow diagrams should include details on equipment brand, model and
version, communication type, intermediary stations of communications, data concentrator and location,
database and location, server details, phasor applications and their latest version number. They should
distinguish between the existing and planned phasor system. The in-depth details on the equipment
source is advocated to be documented with a comprehensive list of PMU settings such as type, number
of phasors, filtering and windowing technique, data rate, PT/CT ratios, calibration and communication.
The primer on synchrophasors is helpful to anyone who is trying to get started on this technology. It is
also helpful if the documentation clearly indicates the group that manages each piece of equipment and
software to help work coordination.

In addition to project documentation, record of ongoing maintenance and troubleshooting procedures is

recommended. Since system problems are often hard to diagnose and resolve a description of
symptoms and problems discovered can be very useful, particularly for a changing staff. Some of this
additional documentation should include:

1. Quick methods to solve common issues.

2. Historical records of equipment settings (especially PMU).
3. Records of issues and problem resolution.
4. Procedures for troubleshooting and problem resolution.

A record of problem resolution should include a detailed description of the observed symptoms
including the starting date/time and related activity in other parts of the phasor system. Once the issue
is tracked down and resolved, the problem details need to be included. Details on the resolution
process and the lead group in the work could be helpful too. If the problem is significant or likely to re-
occur, the technical team solving the issue should prepare a step-by-step procedure on how to
troubleshoot the issue. An index of problem symptoms and resolutions is helpful for expediting the
overall process of troubleshooting. An electronic cross reference is a very good approach for this
indexing.

Page 7
CHAPTER 3. PLANNING, DEVELOPMENT, AND IMPLEMENTATION

3.1 System planning

The first step in building a phasor measurement system is a system plan. Planning should start with the
applications that will use the data and the requirements for data that serves these applications. Data
requirements include but are not limited to:

1. Locations from where measurements are required (PMU location).

2. Signals to be measured at those locations (including phasors, megawatts, MVAR, status or
switch indications, controller values, frequency, rate of change of frequency, etc.).
3. Data rate (rate at which measurements are sent from site).
4. Maximum reporting latency (delay in sending data from site).
5. Reporting reliability (allowable data loss).
6. Measurement characteristics (accuracy, resolution, response time).

It is often the case, however, that the applications that will use the data are not clearly defined or fully
characterized. Since synchrophasor technology is relatively immature, even current applications with
defined requirements are likely to evolve and develop new requirements. Without well-defined
requirements, a recommended approach is to get the most comprehensive measurement set the
project allows in its given scope and budget. Attention should be given to enable overall grid monitoring
and the ability to observe dynamics. Monitoring should include all key transmission inter-connections,
major generating sites, and stations with dynamic control capability, such as direct current (DC)
terminals and Flexible AC Transmission Systems (FACTS) sites. PMUs should meet established
measurement standards. Available communications with sufficient bandwidth is needed and the project
should follow established company practices for system implementation. Table 1 summarizes typical
requirements for some general application types that are in current use. Any particular system may
have different established requirements, so this is only a general guideline.

Table 1 - Typical measurement requirements for general application types

Off-line (studies, Operator Real-time

operation visualization controls
verification, model (graphic Operator (depend on
Parameter validation) displays) alarms State estimation specifications)
PMU location Key substations, Key Depends on Grid coverage by Specialized for
generation & substations, particular bus voltage and control
controller sites coverage of alarm line currents
grid
Measured All voltage and All main grid Selected V & I, All V & I in grid Selected V & I,
signals current (V & I) at voltages, selected other selected other
the substation, selected signals signals
selected other currents

Page 8
 signals
Data rate 20-30/second 10/second 10/second 1/second 10-60/second
(min)
Reporting No requirement, 2 seconds 2 seconds 10 seconds 0.05 - 1 second
latency (max) set by archiving
delay limit
Accuracy High, at least per 1-5% 1-2% 1% 1%
C37.118
Resolution High, .01% Low, 1% Medium, 0.1% High, .01% Medium, 0.1%
Measurement Medium, 50-300 Low, 150- Low, 150-500 Low, 150-500 ms Varies, 16-500
response time ms 500 ms ms ms

In addition to the application and grid considerations, the overall system constraints need to be factored
in. PMUs need to be installed so they have access to the required signals. If the substation is owned by
another utility, a special PMU installation agreement may be needed, or observability may be achieved
from a nearby substation. It may not be possible to install a measurement on a generator, but the line
side of the step-up transformer may be accessible and provide the required information.

Measurements need to be transmitted to the location where they will be used. The first step is from the
substation to a control center. The second step may be to a co-located application or to a higher level
control center. Since phasor data systems operate with a predictable data rate, it is easy to assess the
bandwidth requirements. Some examples are presented in Table 2. Use of a bandwidth calculator
based on the actual implementation for an accurate estimate is recommended. Reliability and latency
need consideration also, but these will usually be dictated by the system architecture and
communications network that is available.

Table 2 – Typical data rate requirements for C37.118.2 format data over TCP in Bits Per Second (BPS)

Transmitted data rate (BPS) 10/s, integer 30/s, integer 30/s, floating point
1 PMU, 2 phasors 8 000 24 000 26 880
1 PMU, 10 phasors 10 560 31 680 38 400
10 PMU, 6 Phasors/PMU average 30 880 92 640 159 840
200 PMU, 6 Phasors/PMU average 486 880 1 460 640 2 804 640

Factors that influence the data rates in Table 2 above:

 Each data report is sent in an individual packet: with a small number of measurements (line
1) the bandwidth is dominated by packet overhead; with a number of measurements (line 4)
the packet overhead is insignificant.
 With larger packet sizes (over 1500 bytes) the packets will be fragmented by the Ethernet
interface, so the data rate will be greater than shown here (~10%).
 Floating point data is 2x the size of integer data (see line 4 where overhead is small).

Page 9
  60/s data is exactly 2x the requirement for 30/s data.

Once the needs for observability (i.e., signals to be measured and their locations) are addressed and
consideration is made for PMU installation and data transmission, the next stage is comparing what is
implemented versus the planned applications, budget, schedule, and other expectations. There may not
be enough budget or time to implement all the expected or desirable parts of the system. Some of the
desired measurement points may be inaccessible. Some of the anticipated applications may not be
available. It is also wise at this point to make a list of possible future expansions.

Several approaches can address budget constraints to cover all the desirable measurements. One is to
use a phased implementation that installs the first PMUs in the most important locations as determined
by system dynamics followed by less significant ones. Another approach is to create a matrix of
applications to prioritize which installations will serve the most applications and the ones of highest
priority. If the focus is on analysis, the system planners can determine the most important locations. If
the focus is on operations, the dispatch and operations group can point out the most important
locations. If there is an ISO or Regional Transmission Owner (RTO) organization directing the work, they
should supply a list that needs the highest priority. Some locations may already have adequate
communications and equipment capable of making phasor measurements. These can be included with
very little budget, even if of lower priority. The most important thing is to create a workable system
within budget that will serve the most important needs of the users.

Various power system monitor, operation, control, and analysis applications are being developed all the
time. Many products take full advantage of the capability provided by phasor measurement. If the
particular application that the user wishes to use is not readily available, in-house development, perhaps
with a university partner, is an option. Another option is to contract with a vendor for modification of
an existing product. There are a number of utilities and vendors involved in this industry that are good
references for advising on new application development. In any case there is more risk developing a
new specialized product than in using well proven existing ones. Synchrophasor technology is relatively
new, and not many people have experience with application of this technology.

When a regional group such as an ISO or RTO is involved, their requirements need to be included in the
TO system planning. The requirements should include mutual agreed plans on identified PMU locations,
measurement characteristics, signal characteristics, communication requirements such as bandwidth
and reliability. Parameters such as data rates, measurement characteristics (M/P class, filtering), single-
phase/positive-sequence, and other characteristics that can be user chosen need to be discussed and
decided. Naming conventions need to be crafted so they satisfy both the TO and regional entity
requirements. The communication method and interconnection issues such as bandwidth, reliability
(including redundancy), point of interconnection(s), latency through TO equipment, and data flow
control need to be agreed. Finally, the longer term issues such as troubleshooting coordination,
operational controls, scheduled and unscheduled maintenance, notifications, equipment calibration,
and replacement policy should be at least be discussed at the planning stage.

Page 10
Planning also includes the overall system considerations such as redundancy, archiving, and cyber
security. Full redundancy should include the PMU, communications from the substation to the TO
control center, all processing within the control center, and communications and systems within the
RTO if that stage is included. It could also include the sensing instrument, such as the PT/CT devices.
Redundancy can be an expensive requirement, so it is advisable to consider the applications and
alternatives. For example, if bus voltages are critical measurements, a particular bus voltage can be
backed up with the measurement of a neighboring bus voltage combined with a line current between
busses, eliminating the need for two measurements on the same bus and the additional communication
requirements. If the application can tolerate short outages, a hot standby or fail-over design could be
used instead of a complete dual system. This document will not present an analysis on the subject, but
rather point out that there are many options for redundant operation and encourage the designer to
consider the requirements of the application and choose an approach that will meet those requirements
within the context and budget of the project.

Most phasor data systems include archiving for the data. Usually only a small percentage of the overall
budget is needed to build an archiving system, so it certainly makes sense to include one, particularly
considering the number of applications that use archived data. A redundant archive system or some
kind of backup is highly recommended. This could be in the form of a fully redundant database system
or a primary archiver with a file writer. In some cases, archiving at the substation is used to provide
back up. The main problem in archiving is failures can happen without much notice and may not be
discovered until someone tries to access the data. Then it is too late. Considering the value of the data
and low cost of archiving, a backup is recommended.

Cyber security is at the forefront of most networking and computer system considerations. It is a
prudent practice and is also in most cases a regulatory requirement. Since phasor systems are heavily
dependent on communications networks, cyber security needs consideration from the first system
designs and through the whole process. The planner is advised to not postpone these considerations.

3.2 System Development - Equipment Selection

The next step is the equipment selection and detailed system and subsystem design. The detailed
design will depend on the particular utility facilities including substation, communication, and control
center equipment. This document makes recommendations about equipment characteristics and leaves
the required selection to a design engineer who has knowledge of utility requirements and facilities.

3.2.1 Phasor Measurement Unit (PMU) selection

There is a wide variety of PMU equipment available. These range from specialized PMUs that have only
a single voltage and a single current input to digital fault recorders (DFR) type equipment that will
handle a dozen or more voltage and current (V & I) signals distributed throughout a substation. The
main PMU differences considered here are:

Page 11
 1. Number of V & I inputs.
2. Distributable input interface units.
3. Auxiliary inputs for analog or Boolean (status) inputs.
4. Timing input.
5. Communication interface.
6. Phasor algorithm options.
7. Accuracy & performance.
8. Multi-function devices vs. single function PMUs.
9. Cost.

All V & I inputs should be 3-phase and the PMU should offer positive sequence as one of the output
options. The number of V & I inputs will vary by the installation and the signal measurement
requirements. Generally the principal bus voltages and major line currents should be included in the
measurement set. Since the incremental expense to include most line currents is small, most utilities
will try to do a complete station measurement. It may be necessary to include several PMUs to cover a
large station.

The designer needs to consider which V & I signals are required, how/where the signals are accessed,
and the types and number of PMUs required to accomplishing this. In some substations, the currents
are only available at distributed relay houses, or even in bay controller boxes. In others, all signals are
brought to a single control house. When choosing a PMU, it is necessary to determine how the signals
are distributed for access. It may require using a PMU that has distributable input modules that can be
located where the input signals are available or using several PMUs distributed where the signals are
located. Data may or may not be collected with a local PDC. It is advisable to consider these issues
before choosing a PMU and other equipment.

Most PMU focus is on the V & I inputs for phasor calculation. In some cases, the user needs auxiliary
information such as breaker or switch status, or control values. For example, dynamics analysis may be
significantly improved by including measurements of DC power order, stabilizers, exciters, harmonics
with percent values, and similar values. Having all this information in a coherent measurement set can
be the key to a forensic analysis or essential to a wide-area control system. Analog and Boolean data is
included in the C37.118 measurement set and is supported by some PMU vendors. If these data types
are required for the phasor system applications, the designer must select units that support them and
with input characteristics that match the installation.

All PMUs require a precise time input. Most will either use a direct Global Positioning System (GPS)
input (from an antenna) or an IRIG-B time code. For the former, the PMU must be located within
allowable cable length distances from an antenna mounting point (including routing into the building).
For substations, this is usually not too difficult if well planned. Fortunately the GPS signal is fairly
resistant to interference from substation equipment. IRIG-B must be likewise provided from a high
accuracy source, usually GPS. The IRIG signal type needs to match the PMU requirements and should
use a high-accuracy modulation (level shift or Modified Manchester). Analog modulated signals do not

Page 12
provide adequate accuracy. The C37.118 (also called 1344) time information profile should be used with
IRIG-B so the PMU can determine when a timing error occurs. IEEE 1588 for timing is being introduced
and should also work well; however there is not enough current implementation to make specific
recommendations.

Most PMUs transfer data over an Ethernet interface and include an Internet Protocol (IP) stack for
network communications. Some models only have RS-232 serial output but can be adapted into a
network using a serial-Ethernet translator. In some cases, serial communication may be preferred due
to cyber security concerns. Both serial communication and Ethernet will handle all communications
from a single PMU up to 60/s data rates. Where data from several PMUs is combined, serial
communication probably won’t have enough bandwidth. Note that the slower clocking rates of serial
communication cause longer delay than Ethernet rates. The designer needs to determine the necessary
bandwidth and allowable delays and design the system accordingly.

PMU algorithms and performance are difficult to assess without testing and a thorough study of the
technology. Generally, a PMU with M-class performance will have more accuracy but more delay in
reporting. A PMU using P-class will have faster reporting but will suffer loss of precision in some
situations. These performances will vary somewhat between devices. Most PMUs offer both classes of
performance. It is recommended that the designer evaluate the intended use of the system and design
accordingly. Choose a few PMUs that meet the expected use and have them tested. Choose PMUs that
pass IEEE C37.118.1 requirements and other particular tests that are specified, and have the best
measurement accuracy. It may be advisable to consult with others through involved technical
organizations for more insights.

A commonly cited difference between PMUs is stand-alone vs. multifunction. The stand-alone PMU is a
hardware device that is built as a PMU and primarily functions as a PMU. A multifunction device is a
device that is built for other functions, such as a relay or a DFR, but includes PMU functionality. An area
of concern is whether the multifunction device will fully perform its PMU functions while performing its
primary task. There is also a concern as to whether hardware optimized for another function can
perform PMU functions as well as hardware designed specifically for PMU functions. Based on
performance testing, a multifunction device is not necessarily better or worse than a stand-alone unit.
The basic hardware configuration of the two types is the same. Both unit types need to be tested under
realistic operating conditions to be sure that internal housekeeping tasks and other functions will not
interfere with the critical PMU functions. The multifunction unit needs additional testing for proper
PMU operation while the non-PMU function is also operating.

Cost is always a consideration, and is sometimes the main driving factor. Generally the overall cost of
installing a PMU with its communication components is 3-10 times that of the PMU itself. Based on this
factor, it makes sense to choose a PMU that fits into the installation with the least amount of auxiliary
equipment. That will often be the cheapest alternative overall.

Page 13
3.2.2 Communication equipment selection

Phasor measurement data is usually sent as packets with phasor, frequency, and other measurements
all corresponding to the time stamp that is included in the packet. This packet is referred to as a “data
frame” since the measurements are more complex than a sampled waveform. The rate is usually 10/s
or greater and the data is pushed from the PMU at the periodic rate rather than polled by command.
The data rate and packet contents are generally pre-planned, so it is easy to calculate the required
bandwidth. The actual equipment will depend on the communication facilities in the substation and the
communication system that connects to other locations. Detailed scenarios are not presented here
since there will be a wide variety of actual equipment needs as well as company procedures and
regulatory policies to follow. The basic requirements of data rate (bandwidth), latency, and reliability
should be presented to communication specialists to design the system and specify the equipment.

3.2.3 Timing equipment selection

The only widely available source of time that is reliable and accurate enough for phasor measurements
is GPS (as of 2013). The GPS uses satellites that transmit precisely timed signals that a receiver
correlates to triangulate position. The satellite time codes are stable and synchronized to universal time
(UTC), so the signals can be used to provide a precise time as well as position. A PMU may have a GPS
receiver installed internally inside or may require a timing signal from an external GPS receiver. The
basic GPS signal is 1575 MHz so it will travel only a limited distance from the antenna. The designer
must consider the PMU location relative to potential antenna locations if the PMU uses an internal GPS
receiver. Note also that a single antenna can supply several receivers by using amplifiers and splitters. If
the PMU is supplied by an external GPS clock, it will receive timing signals by local time code. The most
common local time code is IRIG-B. The level shift and modified Manchester versions have sufficient
precision for synchrophasors. There may be some issues with amplitude and impedance matching, but
these can be readily solved. The IRIG codes were designed for timing on military test ranges, so they do
not have built-in continuous timekeeping capability. In particular, they do not inherently provide
indication of synchronization to a universal time source. These added features can be provided in the
indication field of the IRIG codes (control bits). The C37.118 standards provide a recommended set of
codes tor continuous timekeeping. If the GPS clock can add this code set and the PMU can read it, the
PMU can provide the synchronization time and status required by the standards which are necessary for
reliable measurement system operation. If the GPS receiver is internal to the PMU, the required time
and status can be derived directly. Details for GPS installation and time distribution are left to specific
manufacturer instructions.

3.2.4 Control center equipment selection

Synchrophasor measurements are digital signal representations when they reach the control center.
The applications that use them are signal processing and display programs. Consequently control center
equipment consists of computer processing and display systems, and networking for moving data
between systems. The basic requirements are sufficient computing capability and communications to
support the applications in a timely manner. A key difference between typical data processing and

Page 14
synchrophasor applications is the real-time aspect. Modern servers are very fast and appear to do
things in real-time, but in fact control their own processing priorities according to an internal schedule,
usually optimized for efficiency. Processors using standard operating systems generally are more
efficient running a single task through to completion than switching rapidly between tasks. The normal
preference is to keep efficiency high by minimizing task switching. Synchrophasor data is transferred in
frequent short messages, therefore frequent switching is favored to process the data. Synchrophasor
applications also often need to run on this same frequency processing cycle. If the delays to process
messages are too long, data will be lost. Problems have been encountered using programs not set for
real-time operation, and using virtualized servers with parameters not set for this type of application.
The designer needs to carefully evaluate the real-time requirements of the applications, as well as data
flow since this aspect is different from standard data processing.

Control center applications should include archiving, operation and error logging, system real-time
monitoring, and some kind of visualization. As mentioned before, the incremental cost of archiving is
small but the value of the data for analysis can be very large. A suitable archive also provides a way to
analyze phasor system performance when questions arise. An operation and error log should include all
detected errors in measurements and communications as well as statistics on all aspects of the system
performance. This is essential for problem analysis as well as detecting performance degradation. Real-
time monitoring alerts the operators to failures in a timely manner and emerging problems. Real-time
visualization of the data is available in many forms from simple strip charts to complex analytical and
geographic displays. Some visualization is recommended at a minimum to maintain awareness of the
system’s operation and alert users to the data availability. It is well-established that there is a high value
in using displays to improve operation situational awareness, and it is recommended that visualization
be provided to serve that need.

3.3 System Implementation

System implementation consists of equipment installation, operational checkout, and validation of the
measurements. Typical procedures for similar equipment and systems are generally followed.
Synchrophasor systems are not substantially different than other systems using similar measurements,
communication, and control equipment. A few details are noted here.

3.3.1 Substation equipment installation

A PMU is similar to a relay or DFR in that it needs power to operate, access to V & I signals, and
communications. It also must be connected to an accurate timing source, which is not a necessity for
most other equipment. Equipment power should be DC backed by the station battery so there are no
interruptions during major events. Customary established utility procedures for relays and DFRs should
be followed. The PMU may be primarily a relay or DFR anyway.

Substation to control center communication uses standard communication systems with standard
provisioning and communication checking procedures. It is important to ensure that the allocated

Page 15
bandwidth is sufficient for the PMU data set. Also, it is important to set communication priority of PMU
data relative to other data on the same communications circuit to meet application requirements.
Communication equipment should all be DC powered backed by the station battery as well (don’t allow
an AC modem or Channel Service Unit/ Data Service Unit {CSU/DSU} to sneak in there). Note that NERC
cyber security provisions will apply if the system is used in operations. It is best to be aware of this in
the design cycle and plan it into the installation rather than trying to add it later.

3.3.2 Control center equipment installation

Control center phasor measurement system equipment is basically computer and networking system
components, no different than any other such equipment. It should be installed and configured
accordingly. One additional consideration is the operational priority. Synchrophasor systems typically
handle a lot of data that continuously flows in real-time. The priority for data forwarding and computer
task switching needs to be high enough to prevent losing data. Problems have been observed with
multitasking computers, virtualized servers and inadequate disk storage space where data was lost due
to servicing other tasks. If the planned computer cannot be controlled tightly enough to prevent data
loss, use a separate one. It is much cheaper to buy more hardware than track down data loss problems.
Another added consideration is data timing. If the PDC is doing arrival time calculations, it needs
connection to an accurate timing source, much better than the typical NTP or SNTP time server. This
may require a special GPS module in the server itself or dedicated IEEE 1588 time server. Since a GPS
time input into the control center can be difficult to install, try to arrange the time service in advance.
Computer and networking equipment is typically AC powered, but consider DC backed or a UPS so there
are no interruptions.

3.3.3 Measurement validation

Validation of the measurements is one of the most important parts of the installation procedures. This
consists of validating that the measurement correctly represents the expected quantity and is correctly
reported to the applications in the control center. Some common problems include reporting an
incorrect quantity, reversed signal polarity, incorrect signal scaling or offset, and misidentified signals.
Other less common problems surface as well. If validated initially, the signals will usually continue
unaltered, though things can change during “operation” so it is advisable to have on-line validation to
continuously check for changes.

Phasor measurement initially measures the V & I secondary signals. The relation to primary signals
depends on the PT/CT ratios. These ratios must be entered into the PMU. If the PMU outputs the
phasor estimates in floating point format, they are scaled to primary values using the PT/CT ratios. If
the PMU outputs the phasor estimates in integer format, the ratios are included in the scaling factor
included in the configuration message. Older PMUs may require some external scaling. The user needs
to apply the right ratios and is responsible to update them if they are changed.

The PMU sends data to the control center equipment in packets where the data is identified by the data
type and order in the packet. A configuration message provides signal names and data type, and their
mapping to the data. If the configuration is incorrect or decoded in error, the mapping may be wrong

Page 16
and signals may be corrupted. Also, the configuration may change at the sending end, but the change
may not be detected at the receiving end, also creating errors. Some errors in decoding are detectable
in the data, some are not. Where possible, most phasor system processing equipment (usually a PDC)
will detect and keep the configuration synchronized with the data. Where the error is not obvious, the
problem will only be found through data examination. Periodic examination of all data is recommended
(see maintenance section).

The principal process for initial validation consists of comparing the actual signals with the
measurements and at successively higher levels. First, the PMU measurements are compared with
values measured in the substation using test instruments or installed meters there. With accurate and
full test instrumentation, the measurements can be validated to 1%. Substation validation also requires
checking the timing operation.

At the next level, the measurements are validated against SCADA or other reported instrumentation.
This level of validation is more for checking the signal reporting, flow magnitude and direction, and
overall reporting than precision. Comparisons at higher precision can be made by comparing the
reported measurements with a state estimate during steady state conditions (a time when there is little
dynamic activity). High precision will not be obtained from every comparison since the power system
varies constantly even at steady state, and the dynamic responses of the two measurement systems are
different.

The final level of comparison is at an ISO or RTO which receives measurements from many TOs. At this
level, comparisons with regional phase differences will be evident. Other than that, it is basically the
same as at the TO control center.

Appendix A is a detailed procedure for this validation process from the substation to the control center.
It discusses the measurements to be made at each stage as well as the process. Validation can be done
at the control center level without doing the substation level, though it is possible some problems may
be difficult to resolve without investigation at the substation. The control center validation process is
basically the same whether done at the TO or ISO level. Refer to Appendix A for more detail.

Page 17
CHAPTER 4. OPERATION & MAINTENANCE OF PHASOR SYSTEMS

4.1 Operation and analysis

Most user interactions with phasor measurement systems will be through vendor provided applications.
These applications should provide data validation functions including appropriate indications where the
measurement may not be reliable. This section provides key points that the user should confirm to be
sure the application has the indications it should provide.

Lost data usually requires some kind of recognizable placeholder in the message or data store. It will
usually be indicated by NaN, -9999, or a separate flag. The display, analysis, or other package needs to
indicate when the data is missing rather than just offering the placeholder as though it is actual data.
Some PDCs will repeat old values to fill in missing data. This needs to be flagged and indicated as well.
(The standard IEEE C37.118.2 designates bit 9 in the status for this purpose).

Each data sample has a status that indicates if there is a PMU, sync, or time stamp problem, or the data
is simply in error. The application should be able to indicate when these problems occur and flag the
data appropriately so the data will not give the user false indications.

Phase angle measurement is based on UTC time supplied to the PMU. The measurement Status bit 13
indicates if the PMU time was UTC synchronized when the measurement was made. The state of
synchronization cannot be determined from the phasor value itself. The user needs to check the sync
flag before using phase angles.

The frequency measurement is usually derived from the phasor values. If the input that is providing the
phasor fails, the phasor will not be accurate, and the frequency measurement is likely to be bad as well.
The application should be able to link the frequency to a source signal and thereby be able to indicate
when the frequency should not be used.

Other data impairments are detectable using other more complex techniques. These will be explored in
the next report that includes a validation and conditioning algorithm. At this point, it is recommended
that the user examine the validation techniques built into each application and compare them with
input processing to be sure they are compatible. Most data problems can be resolved by careful
implementation using techniques and flags that have already been established.

4.1.1 On-line data validation

On-line data validation requires building data checks into the processing applications with conversions
to displays and operations. There are three essential functions here: indicate data problems to users,
provide real-time system problem alarms, and create a record of data problems. There are three areas
that need to be addressed:

1. Indication on displays and in alarm functions that are using the data that the data may be
impaired. This will prevent users from taking incorrect actions.

Page 18
 2. A separate alarm function to tell users where a problem has been detected, what it is, and
whether it is still active.
3. System or process to record data issues to evaluate performance and find problems that are
developing. A record of performance is also helpful for analyzing intermittent problems.

Data usually comes into a PDC which aligns it with other data and then distributes the data to other
PDCs and applications. It is probably the best place for both real-time alarming and creating a record of
data problems. It could create a special flag for downstream applications, but that flag will have to be
blended into the data stream and the applications configured to look for them. It is recommended that
the PDC build indications into the data itself so special processing is not required, and have the
applications do validation indications appropriate to their tasks. This will maintain interoperability of
systems and programs.

4.1.2 Off-line data validation

Off-line validation is much easier to do since there are no time constraints. It also does not require
alarming or error and performance recording. It is recommended that on-line detection and flagging
information be retained in the data set so this does not have to be done again. It also saves time when
the user does not have to re-examine all aspects of the data.

Off-line processing may be performed manually or automated. Automated processing such as for
baselining studies or anomaly detection requires techniques to detect the parameters under
investigation, as well as validate the data that is being used. These special processing applications will
often include data patching, down-sampling, and other modification processes specifically to suit the
objectives. These applications should also provide flagging to alert the user when a data anomaly is
detected that is not expected. No other recommendations are offered here since these specialized
applications are too diverse to attempt to cover.

Manual off-line processing is the most common. This includes routine examination of events,
determining operating parameters, model validation, and other related work. One of the most common
issues is that most manual processing does not automatically alert the user to data problems that are
indicated by the status and other flags. Unless the user is educated to look at the status and takes time
to do so, invalid data may be displayed as though valid, leading to incorrect conclusions. Analysis
programs should be equipped to read data flags and indicate problems to alert the user.

Other than the status flags and other indications embedded in the data, validation should include simply
observing unusual traces, patterns that don’t make sense, and comparisons with other measurements.
Some unusual patterns include:

 Flat line – continuous line with no changes at all indicating the variable being reported is not
being updated. This could be at the PMU or at a PDC that is replacing lost data.
 Noisy data – could be higher frequency oscillations. Expand time scale to see if an oscillatory
pattern appears. Measurement could be near 0 so it is just noise on signal input. With no signal
input, a phasor angle will be random noise.

Page 19
  Unusual pattern – signal is following a perceptible pattern that does not look like a simple
oscillation. It could be caused by a clock that is resetting on a regular basis, a stuck
measurement bit, an algorithm that is failing at a regular place, and so on. Nearly anything is
possible, so the only way to resolve these issues is careful examination and comparison with
other signals.
 Measurements that don’t make sense – a voltage way too high or low, an impossible power
flow, phase angles that are unlikely, etc. These problems are usually scaling errors, but also
include an incorrectly identified measurement (i.e., the wrong signal), a direct measurement
error, data patching errors, bad timing input to PMU, and similar issues.

The next level of manual data validation is comparison with other measurements. SCADA is usually the
most available. This data is reported at a much slower rate, so it will only confirm readings within that
rate. Steady-state measurement levels can be confirmed. Phasor data can be visually estimated at a
slower rate, but more precise comparison requires an averaging type of down-sampling. Voltage
measurements and power flow computed from phasor data should match reasonably close.
Discrepancies need to be investigated and recorded.

State estimation is required for finding bus voltage phase angles from SCADA data. A state estimate
with low residual error can be used to confirm phasor values from the same point in time. As long as
there are no major system changes occurring at that time, the comparison should be close.

For dynamic measurement validation, DFR recordings are a good source. They have been used in a
number of cases to determine whether the PMU data indications are valid. DFRs usually record data at
a sample rate ranging from 960 to 6000 samples/second. The file lengths are usually only a few
seconds, so validations can only be made for very specific measurement points. Phasors show the
envelope of the 60 Hz sinusoid while the DFR shows the entire sinusoid. While some simple
comparisons can be made by visual observation, detailed analysis requires some conversion. One
approach is to estimate synchrophasor values directly from the DFR samples. If their time stamps are
accurate enough, direct comparison can be made. If not, the traces can be lined up manually to correct
the time offset after which they can be compared. Another approach is to look at the signal spectrum of
each record. The DFR data has a much higher frequency spectrum due to the higher sample rate, and
can be used to confirm frequencies detected by the PMU. It will show if a signal detected by a PMU is
present, absent, or possibly aliased from a higher frequency. Some DFRs also include event recording
inputs. Some utilities use a separate sequence of events (SER) recorder for this purpose. If the PMU
also records Boolean information, these devices can be used to validate the PMU data. Note that the
time stamp is precise for the phasor and frequency values, but Boolean data may represent a digital
signal (change of state) that occurred between time stamps. When a Boolean value changes state, the
new value will be reported with the next phasor report so that it appears to occur at the time stamp,
even though it actually occurred between time stamps. So the time stamps for events reported by the
DFR and the PMU may differ, but no more than the interval between PMU reports.

Page 20
4.1.3 Data analysis in operation

Regular analysis of events can significantly enhance system operations. Phasor data will show dynamic
activity that is not apparent from other measurement sources. It also provides a good check on other
system alarms and measurements. In particular it is recommended that:

 Alarms for saving phasor data be set so that event records are easy to locate and not
overwritten
 A policy is established to examining system responses to certain events, such as an event that
causes a frequency excursion over 100 mHz, voltage sags below 5%, generator trips at > 250
MW, etc.
 Alarms based on synchrophasor data are established to alert operators for unstable conditions
such as low voltages, excessive system phase angles, low frequency, etc.
 Operating procedures are established based on synchrophasor system alarms.

These procedures can significantly enhance operations and planning by including full consideration of
dynamics as well as wider measurement area in most cases. Regular examination of phasor data also
supports monitoring measurement system performance, helping to spot problems before they lead to
greater interruptions.

4.2 System maintenance

Synchrophasor systems are relatively new and experience with their maintenance is limited. Most
utilities plan to use the same maintenance schedules already established for similar types of equipment.
These recommendations are limited to aspects that can be estimated or predicted by common sense.

4.2.1 Hardware maintenance

PMU equipment in the substation and hardware PDCs are the only actual equipment that are specifically
for a synchrophasor system. Measurement accuracy depends on the PMU and periodic calibration is
needed to assure precision. There are a minimum number of components that will affect accuracy in a
PMU. In most PMUS there is an analog input circuit leading to an A/D converter. Once changed into
digital form, the value is handled mathematically and will not degrade with time. Only changes made in
the analog circuitry and converter can degrade the measurements and cause changes in measurement
over time. A well-designed system should maintain calibration for several years. Since the functional
application is similar to relays, DFRs, and meters, it is suggested that routine calibration be initially
planned on the same interval as that for these devices. In the survey, several utilities planned a 5-year
maintenance interval. It is recommended to plan the first calibration interval no longer than this. If the
readings are observed to change more than the expected accuracy level (1%) during the planned
calibration interval, it should be shortened to maintain the accuracy.

Page 21
System components such as routers, modems, multiplexers, and computers should be maintained on
the company established schedule and policy. No particular recommendations are indicated relating to
their use in a phasor measurement system.

Applications and most PDCs operate on standard personal computer (PC) hardware and operating
systems (Windows or Linux). Hardware PDCs are uncommon, but operate on some kind of computer
type hardware and an Operating System (OS). In all cases, the accuracy does not degrade over time, so
no calibration is needed. What is needed, however, are patches and upgrades for the programs and OS.
These should be applied in a timely manner and documented.

4.2.2 Performance maintenance

Measurements from the synchrophasor system should be compared regularly against similar
measurements from other systems, such as metering, SCADA and state estimation. The best practice
would be to do these comparisons continuously; however if that is not possible an annual comparison of
most readings is highly recommended. This will detect problems that have come up, such as
undocumented signal changes, PT/CT errors, failed A/D units, etc. This kind of change is outside the
scope of normal calibration, but happens and is sometimes not noticed until a problem comes up. (It is
also a good way to find SCADA failures.) These comparisons should be documented in such a way that
slower degradation can be detected as well.

Resource use needs to be checked and documented on a regular basis. This includes communication
bandwidth used for the phasor system and disk space for data storage. Changes in the monitoring can
add more data and impact the available bandwidth without obvious indications. Also, many
communication channels are shared, so growth in other systems would impact the phasor system. On
the data storage side, data growth is normally planned with some allowance for changes. However
there are often unplanned expansions such as extra events and user analysis records that significantly
increase the space requirement. Disk space should be checked regularly to be sure there is sufficient
space to avoid loss of data due to disk overflow. Communication bandwidth should be checked at least
annually and disc space at least quarterly.

Maintenance procedures and intervals are established by the utility that operates the equipment in
accordance with regulatory and other controlling party requirements (e.g., ISO). The recommendations
given here are based on observed practices and problems, and should be considered as advisory to
establishment of a company plan.

4.2.3 Configuration management

System configuration includes measurement sub-system identification and characteristics,

communication sub-system connections and routing, and application sub-system settings. These
configurations are done initially as part of the installation and checked in the validation process. As long
as the system doesn’t change, it should all operate as first planned. However, change is inevitable and a
process for configuration management is critically important. Changes in configuration need to be
planned by the person or group that manages that particular aspect. For example, an IT department

Page 22
usually manages routing and firewall configurations in the network. They should plan all such changes.
After change planning, the other stakeholders need to be notified of the planned changes, as these
changes may affect other parts of the system in unexpected ways. For example, the substations group
may be adding a few measurement points, but that may increase the packet size enough to require
additional communication bandwidth. Finally, after notification and sufficient time for a response, the
lead part can proceed with the change. This should assure a smooth transition to the new configuration.

Page 23
APPENDIX A. PMU AND SYSTEM INSTALLATION AND VALIDATION PROCEDURES

A.1 Introduction

This document provides the requirements for validation and documentation of phasor system
installation. It is broken into sections for each level of installation. A.2 describes PMU certification and
calibration requirements. A.3 focuses on the PMU installation in the substation. A.4 confirms the data
as reported to the control center. It should be followed in order to identify problems where they first
occur.

Test equipment and methods may be called for that are beyond the practice of a particular utility or
staff function. Signal readings may be indicated that do not exist or are inaccessible. The procedure
attempts to offer alternative processes, but in places may not do so. The reader is advised to use
alternatives that match the intended process as closely as possible.

A.2 PMU compliance with C37.118.1 and calibration

Compliance with C37.118.1 should be shown. The utility, vendor, or a third party can perform the
required tests. The requirements in the standard cover both overall performance and calibration.
Compliance needs to be shown for the particular PMU with the specific settings that will be used, since
differences in settings can significantly change performance. A complete set of tests would only be
needed on a single unit, since most of the processing is a software function and these will not change
from unit to unit (unless the actual design is faulty). Differences between units of the same type and the
same settings should only be found in input scaling and phase angle measurement. These differences
are determined by calibration testing.

The measurement accuracy capability of a typical PMU is in the 0.5% range. Calibration requires signals
and measurement devices that have resolution and accuracy in the 0.05 to 0.1 % range. It is difficult to
do this in a substation setting, so this type of testing needs to be performed in a laboratory. The basic
C37.118.1 performance requirements require accuracy in the 1% range, but the test results can be used
to as high accuracy as the test equipment will allow. The C37.118 steady state magnitude and angle
tests can be used for calibration.

For best operational accuracy, every PMU should be calibrated in the lab before deployment in the field.
The scale factors should be adjusted based on these calibration results to give the best accuracy
possible. A test record needs to be kept, and periodic recalibration is required.

Since individual lab calibration may not be practical, a generally acceptable approach is to sample test a
batch of PMUs. If they meet the requirements consistently, testing of a subset of deployed PMUs
should be sufficient. Examination of the results may show a systematic error in the measurement that
can be improved by adjustment of the default scaling. However, in general, calibration adjustments for
individual units cannot be made from a sample test.

Page A-1
A third alternative requires manufacturer certification and calibration. This can include calibration with
adjusted scale factors. The utility should spot test a few units to be sure calibration is good.

A.3 Substation installation requirements

The phasor system substation installation consists primarily of installing the PMU with its related input
and output. It will have V & I inputs for measurement, a time input for synchronization, and a data
output. It may have additional measurements of status (Boolean values) and analog values, either input
through an A/D or internally calculated. Validation of this installation consists of assuring that the inputs
are correctly connected, quantities are correctly identified (labeled), the scaling is correct, and the data
I/O is operational. Calibration to the system requires local test instrumentation.

A.3.1 Check that the clock used for synchronization to UTC is on time and locked on time. Check that
the PMU correctly indicates when time is locked and that the lock is steady.

 Method

The clock may be a GPS receiver internal to the PMU or may be a substation clock that supplies time
using IRIG-B or another format. The PMU may provide a display to show the clock time as received by
the PMU or allow using a program to look at the data. Alternatively use a data reader to look at data
and time. A PDC on a laptop is another way to locally check the data output. UTC time does not change
with seasons for daylight. It has a constant offset from local standard times. The offset from Eastern
Standard Time (EST) is -5 hours (subtract 5 hours from UTC to get EST).

Lock should be indicated by the receiver, whether internal or external to the PMU. It is required to be
indicated by the output data. If there is no user indication on the substation equipment, start output
data and observe the indication there using a local device as indicated above.

To determine that the lock indication reports correctly, remove the time synchronization source from
the clock. The PMU should report sync is lost within 1 minute. It should likewise report sync is regained
when the time is reacquired (the receiver may take a few minutes to re-lock after the antenna is
reattached). Note that if GPS is used as the prime source, this exercise will involve disconnecting the
GPS input from the clock. If the clock is internal to the PMU, disconnect the GPS input. If the clock is
external to the PMU, the GPS must be removed from the clock, not just the time signal (i.e., IRIG-B) from
the PMU. The PMU must determine that the time signal is synchronized, not simply that it is receiving a
time signal (the PMU may have a valid time signal but the signal may not be synchronized to UTC).

The lock to GPS must be continuous. In reality the receiver may have short unlocks while switching
satellites, but the receiver should ride through those with its internal oscillator. If there are any
significant reception difficulties, they should show up within a day. Monitor the lock signal for a 24-hour
period to be sure that there are no dropouts. Any dropouts require investigation, remediation, and
retest. The clock or PMU may have a built in monitor which can be used to confirm lock. Alternatively

Page A-2
collect output data using Connection Tester or a PDC to analyze for dropouts. A PDC should provide
performance statistics that will show the sync performance.

A.3.2 Confirm that the phasor measurement magnitudes are within 1% of input levels. Also confirm
voltages are within 1% + 1kV and currents within 1% + 100A of comparable substation measurements.

Traditional substation measurements are bus voltage and power flow in lines. Some substations will
provide metering or other indications of line currents. PMUs measure V & I. The challenges in
measurement comparison are that the values vary constantly, so the user has to estimate at a moment
where the values are not varying rapidly and input and output can be captured at the same time.

Substation values can be read from test equipment, substation meters, or local SCADA systems. PMU
values are often provided by a PMU display. Alternatively the PMU vendor may provide a viewing
program for a PC, or Connection Tester or a portable PDC can provide the data.

 Comparison of phasors with actual inputs

Connect test equipment to the PMU V & I inputs. The test equipment should be within its certification
period and have an accuracy rating of 0.1% or better. Compare the input values with the values
provided from the PMU. Make the readings for comparison at as close to the same time as possible to
eliminate variations. Make the comparisons by phase, i.e., Va input to Va phasor, etc. Be sure to apply
sqrt(3) corrections where needed. Note that generally PMU voltage inputs are phase-neutral and
current inputs are phase, so Y-Δ scaling is probably not needed. The comparisons should be within 1%,
which is the minimum C37.118 requirement. If the error is larger, investigation should be done as there
is probably a connection or setup error. If PMU also reports positive sequence, it is worth checking by
averaging three single phase readings for comparison (the phase angles should be close enough).
Positive sequence thus tested should be within 2%. This additional comparison assures that the phasing
is properly connected.

 Comparison of phasors with substation instrumentation

This will be less accurate than the first step since installed substation instrumentation is usually less
accurate than portable test equipment and is often connected to different PT and CT signals. It does
provide local validation that the PMU is correctly connected to the indicated PT and CT signals as
designed. There will usually be voltage measurements in the substation and sometimes current
measurement. Compare individual voltage and current phases as above. Try to choose voltage readings
from the same PT for comparison. If there are single readings, just compare the appropriate phase.

If there are only power measurements on feeders, compare with power computed from V & I phasor
values. A tool that will provide these readings on-line is the best approach. If no such tool is available,
power can be computed manually = + using rectangular components or
= cos(∅ − ∅ ) using polar components.

Page A-3
The main point to these comparisons is to spot wiring, naming, and scaling errors. For example, if the
wrong current is wired into the PMU, the computed value will probably differ considerably. The
comparison needs to be only to an accuracy that will reveal errors since the accuracy is checked directly
in the test above. A comparison of 3-4% is probably sufficient, allowing for 1% in the PMU and 2-3% for
the local measurement.

A.3.3 Confirm that the phasor measurement angle differences are within 1-3 degrees of corresponding
input signal angle differences. Also confirm the angles with comparable measurements in the
substation.

Phasor angle measurements are made relative to UTC time and there is no direct way to validate these
values short of using another PMU. However the phase angle between signals can be readily
determined by a number of methods. As above, portable instrumentation is prescribed here for
calibration and using installed substation equipment for signal validation. The angle should be validated
to within 1 degree, but this may not be possible in all test situations. The test staff needs to use
judgment as to whether the measured errors correctly represent the PMU errors or whether they are a
result of imprecise test equipment and methods.

 Comparison of phasors with actual inputs

Choose a reference signal (e.g., the bus voltage) against which all other angles will be measured.
Connect phase angle measurement equipment to the reference signal and another input signal (voltage
or current). Compare the angle between the same signals measured by the PMU. Make the readings
for comparison at as close time wise as possible to eliminate variations. This comparison is easiest with
a phasor reading device that allows setting a reference that will be subtracted from the other signals to
directly display the angle difference. Alternatively, the process can be done manually. Choose a time
when the system frequency is very close to 60 Hz as the phase angles will vary more slowly. Compare
measurements of the same signal and phase. The input and PMU measurement should be within 1
degree of each other. PMU qualification based on 1% Total Vector Error (TVE) requires phase angle
accuracy 0.57 degrees or better. Testing has shown that most PMUs measure phase angle to accuracy
near .02 degrees. Test equipment should be capable of 0.1 degree accuracy, but check the
specifications. If the current is low (under 20% of rated current) the measurement may be less accurate.

 Comparison of phasors with substation instrumentation

Compare PMU measurements with local installed instrumentation. Traditional measurement

equipment does not measure phase angle. Angle between V & I can be determined from real and
reactive power measurement. Phase angle is defined as φ = tan-1(Q/P) where Q and P are reactive and
real power respectively. The phase angle for each current relative to the related voltage can be
determined this way. These standard instruments will not give the angle between different bus voltages
or currents for which there is no corresponding voltage. As with the specialized instrumentation, when
the current is low, the watt, var, and phasor measurement will be less accurate. Allow larger errors for

Page A-4
currents less than 20% of rated current, and do not expect to make comparisons with currents less than
10% of full load.

As above, this step is intended to spot wiring and naming errors (scaling errors do not affect phase
angle), so compare measurements with this in mind. The level of comparison should be enough to
confirm that the correct signal with the correct phasing has been selected.

A.3.4. Confirm that analog measurements are within 5% of other measuring devices in the substation

Synchrophasor data communication can include samples of analog signals that are not phasors. The
C37.118 protocol includes this data type as “analog” data. It can be represented in integer or floating
point values. The standard does not specify what these analog signals are, so they can be any
continuously varying signals such as power (watt and var), control signals, local readings (pressure,
temperature) or any other continuously varying signal. They can be sampled from an external source or
produced internally in the PMU from other inputs. The standard also does not specify how the
measurement is to be made, scaled, recorded, or any other measurement parameter. The data type
was added to include measurements in the synchronously sampled data set that are important to using
phasor system data and adding understanding of power system operation. Some of these values may
have local reporting in the substation and some may not. Since there is no standard specification for
making these measurements and what they represent, this confirmation test is simply a best effort to be
sure that the reported data reasonably represents the signal it is supposed to.

 Method

In cases where there is no means to locally observe the measured quantity, this test can be ignored.
Where it is observable, find a suitable means to read the values as sent from the PMU. These values are
often provided by a PMU display. Alternatively, the PMU vendor may provide a viewing program for a
PC, or Connection Tester or a portable PDC can provide the means to read the data. Locate a
measurement display for the quantity in the substation. In some cases portable test instrumentation
may be required. If computed internally, the user can compare PMU input values or even use the PMU
phasor values. Compare the two values. The values generally should be within 5% of each other,
however, in some cases it can be much closer and in some other cases not as close. It is left to the
user’s discretion to determine an appropriate level of correspondence and validate the measurement to
this level. Deviation of accuracy acceptance from the given 5% minimum needs to be documented.

A.3.5. Confirm that digital status measurements report the correct Boolean state

Synchrophasor data communication can include digital status indications. The C37.118 protocol
includes this data type as “digital” data. Digital status is represented as a Boolean 0 or 1 binary value.
The standard carries Boolean values in blocks of 16 as 16-bit words. The standard includes the ability to
specify which bits in each word are valid representations (that is, in use) and what is the normal state (as
opposed to an alarm state). These status can be used for any binary signal (i.e., on or off) such as
alarms, switch position, position, etc. The user must assign suitable identification and determine how
the indication will be used. The data type is included so the user can include indications in the

Page A-5
synchronously sampled data set that are important to using phasor system data and adding
understanding of power system operation. Some of these values may have local reporting in the
substation and some may not, however it should be possible to determine the local state of all such
indications.

 Method

Find a suitable means to read the values as sent from the PMU. These values are often provided by a
PMU display. Alternatively the PMU vendor may provide a viewing program for a PC, or Connection
Tester or a portable PDC can provide the means to read the data. Locate the source of the signal and if
possible, a display or indicator in the substation. The goal is to determine whether the input state is
correctly indicated by the phasor data. Observe that the output correctly indicates the input. Change
the state of the input and observe that the indication changes. In cases where the source cannot be
changed, such as breaker position on a critical line (that cannot be taken out of service), operate the
input in a test mode to be sure the indication changes. It is important to actually change the input to
the PMU (e.g., voltage on the PMU input) to be sure that the actual PMU input is working, it is mapped
to the correct data, and voltage or current input thresholds are satisfied. Do this for all digital inputs
that are sent as digital status.

A.3.6 Substation installation confirmation documentation

Documentation in the form of notes and test results needs to be provided to show each test was
successfully completed. It should include a description of the test methods, equipment used for tests,
raw measured results from the instruments and the PMU, calculated values, calculation methods and
formulas, and the comparison results. Anything that was observed which could be of interest should be
noted as well, whether it is relevant to the given test or not. Any repairs or changes made as a result of
these tests should also be noted.

A.4 Control center installation requirements

The control center installation includes the PDC, communication systems, and various applications. This
checkout is primarily to confirm that the communication to the PMUs is operating correctly, the data is
correctly identified and scaled, and interstation phasing is correct. It also provides an initial comparison
with state estimation to confirm if the overall measurement schemes compare.

A.4.1 Confirm that received data match the setup and the time stamps are within 3-seconds of the
local time

There will generally be a number of remote devices (PMUs and PDCs) reporting to the control center.
The data received from each device needs to be validated that it matches the data description. At this
stage only the message reports need to be validated; data content will be validated in A.4.3. If the time

Page A-6
stamps do not match the actual time, the data will not match and cannot be combined. These issues
need to be resolved before the data itself can be validated.

 Method

Data is mapped as binary values into a frame (message) that is sent from the PMU to the control center.
These frames will vary in size and content from station to station. The receiving device, usually a PDC,
will make the connection to the remote and request a configuration message. This message will provide
the data description including names, scale factors, data type, and the location in the data frame. The
PDC uses this information to decode and scale the data to usable values. The PDC should do this
automatically, but it is highly recommended that the test engineers examine the values. In particular,
compare the signal names and scaling with a listing from the design documents. The names should
match, all data items should be accounted, and scale factors should seem reasonable. Data values will
be more closely examined in A.4.3, but a review at this stage can reveal problems that are more difficult
to diagnose later.

A timestamp is included with all data frames. It indicates the time of measurement. Data frames should
be sent with minimal delay, so they should be received with a time stamp that is no more than a few
seconds earlier than the current time. The data time stamp is in UTC time which will be offset from the
local time and does not change for daylight seasons. If a local time is used, determine what the current
offset from local to UTC time is. The offset from Eastern Standard Time (EST) is -5 hours (subtract 5
hours from UTC to get EST), and Eastern Daylight Time during the summer season -4 hours. Use an
accurate computer time reference and visually compare the received data time with the computer time.
This can be done with some PDCs or alternatively a data viewing program for a PC, or Connection Tester.
A visual comparison can determine if the times are within one second of each other. If the received
data time appears to be more than one second ahead or more than 3 seconds behind local time, there is
likely a clock error at the PMU or some other setting problem that needs to be resolved.

A.4.2 Catalog communication errors and show that the overall data loss is less than 0.1% over a 24-
hour period

Each remote device reporting to the control center has a communication channel which may include a
number of links, translations, carrier systems, and so on. Each element of the communication chain can
cause impairments of one type or another. This test simply observes and documents these impairments
over a 24-hour period, as well as assures that the overall data loss is less than 0.1%. Phasor data
systems should operate at loss rates much less than 0.1% (which is approximately loss of 2 data
samples/min at 30/s data rate), which will generally provide sufficient data availability for most
applications. A well-designed and implemented system can be expected to have a loss rate <0.001% (1
frame/hour) during normal operation.

 Method

For each remote measurement input to the control center, monitor the reception of data over a 24-hour
period and record all data loss. The inputs will usually be from a single PMU at a substation, but can be

Page A-7
from a substation PDC or possibly from another PDC that collects data from other PMUs. The record of
data loss should include separate tallies of data that is received but corrupted, data that is not received
but expected, and any unexpected changes in data format. The receiving device needs to have a long
enough wait that it includes all data received up to a reasonable delay of at least 10-seconds (i.e., it
should count data that is up to 10-seconds delayed as successfully received). Data that is received with
a time stamp more than 10-seconds from the current time should be counted as lost with a time error.
The tally should also include data received but flagged as an error. These flags are data invalid, PMU
error, PMU sync lost, or sort-by-arrival. For data loss that is significant (> 0.001%), the record should
include sufficient information to determine the pattern of data loss, such as all data loss in one long
dropout, data loss of 1-2 frames at a time spread over the time period, a regular repetition for dropped
frames, etc. For this requirement, the received data itself should be saved so that a more detailed
analysis can be performed.

 Special documentation

Documentation for this test should demonstrate that the data loss for each input channel is < 0.1% over
the 24-hour test period. It should include a list showing the number of times and overall percent of
each category of data loss including:

1. Data not received.

2. Data received but corrupted (Cyclic Redundancy Check {CRC} error or similar).
3. Data received with time stamp error.
4. Data received with error flag set
a. Data invalid
b. PMU error
c. Sync lost
d. Sort-by-arrival

In cases where data loss is significant (> 0.001%), the store of data captured during the test should be
further examined for the pattern of loss; this will be included in the report. Describe the method for
performing these tests and list of the equipment used.

A.4.3 Compare received signals with SCADA measurements and confirm they report the same values
within the following limits:
1. Voltages – 1% + 1kV magnitude.
2. Currents – 3% + 1% full-load current magnitude & same direction.
3. MW & MVAR – 5% magnitude & same direction.
4. Frequency – 0.003 Hz.

SCADA provides fairly complete power system measurement capability for most utilities. This step
confirms that the phasor system reporting is consistent with that provided by SCADA. Since SCADA is
the established system, it will generally be used as the benchmark or reference that other

Page A-8
measurements will be compared with; if they are reasonably close to the SCADA measurement, they will
be accepted as correct. This comparison will first help to locate measurement quantities that are
mislabeled or otherwise incorrectly identified. Secondly, it will enable finding scaling or wiring errors.
Lastly, it may result in correction of SCADA measurements. Some measurements from phasors will not
have a SCADA equivalent, so can be skipped in this procedure. These measurement comparisons are
intended to be taken once and reported; if taken several times over different operating conditions, the
results will be more accurate and useful, but this is not required.

 Method

Comparisons can be done with live displays side by side, or snapshot data samples. The important point
is that data from each will need to be compared at the same time.

Both systems provide voltage measurements. As long as the comparisons are from line or busses that
are connected, the voltages will be essentially the same. In many cases the SCADA measurement is a
single phase. If the phasor single phase measurements are provided, the same phase can be used for
comparison. The positive sequence is generally close enough. Phasors generally report line-neutral, so
use appropriate conversions.

SCADA systems often do not report currents. If they are available, make phase by phase comparison
where possible. Note that the current direction is arbitrary and can be wired with either direction
representing positive. Most systems designate the current as positive when it is out of the bus as
indicated by being approximately in phase with the voltage. For consistency, it is recommended that the
current direction be designated the same as in the SCADA system.

SCADA systems usually report MW and MVAR. Phasor systems usually do not, but they are easily
calculated from the V & I measurements. An additional application may be required to make these
conversions, or the user can manually calculate them from phasor V & I values. Once calculated,
compare SCADA and phasor system MW and MVAR values. They should be within 5% of each other.
The direction indications follow the current direction choices. If the directions do not match, the
problem is probably the designated current direction.

An important point to consider for current, MW and MVAR calculations is the reduced precision at low
current levels. If current is less than 20% of the indicated full scale, the measurement accuracy is likely
to be less than the rated accuracy, and the resulting comparisons will be degraded. The error is usually
greater for the angle, so computing MW or MVAR is likely to have high error. For example, a line scaled
for 2000 amps but with only 100 amps of load (5% rated) might show a VAR reading of 20 MVAR when it
really only has 3 MVAR. The point here is not to expect to get good comparisons with this kind of
reading. It is recommended to perform comparisons when the line has higher loading, at least 30% of
capacity. However it is good to record results at different levels for comparison. The flow direction will
usually be ok even if the reading is low. Lines out of service will often provide measurements, but they
tend to be very random. Be careful to limit comparisons to lines in service.

Page A-9
Frequency varies constantly but generally in a limited range during normal steady-state operation.
Choose a time when frequency is changing slowly to make the frequency comparisons between PMUs
and SCADA. It is best to compare the frequency from several PMUs against each other to spot any that
are in error. Pick a few PMUs that are in close agreement, average their readings over 1-3 seconds, and
compare with that from SCADA. Keep in mind that the SCADA measurement probably averages
frequency over a second or more. If the phasor measured frequencies are close to each other but
different from SCADA, then the SCADA measurement is probably in error.

A.4.4 Validate data status indications

Every frame of data includes a 4-bit data quality status indication. These 4-bits are data valid, PMU
error, sync valid, and sort by arrival. The “good” or normal state is cleared to 0; when set to 1, the bit
indicates the alarm or abnormal state. These bits are set by the measurement device (PMU) or other
processing devices (PDC) based on measurement conditions or errors detected in further data
processing. Appendix B describes these bits and their use. This step is to validate that these quality
indications are being properly received and handled by the control center equipment.

 Method

For each incoming data stream, observe the status indications with a device that receives the incoming
data. Confirm that all bits are cleared (to a 0) indicating that all incoming status indications are good. If
any are set to 1 indicating a trouble condition, confirm that the impaired state is actually valid. No
further testing is required for this requirement.

While not required, further testing of system operation is recommended. First, set up a data output
from the control center PDC. Observe that the input under test is included in the output. Disconnect or
disable that input only. Observe that the output data for that input is marked as invalid data. Secondly,
if the PDC allows manually setting the input to sort by arrival, set it that way. Observe that the output
data stream indicates the data is timed using sort by arrival. Thirdly, if assistance is available at the
substation, have the timing synchronization removed from the clock (remove GPS antenna) so the clock
will go to an unsynchronized status. Observe that the sync bit is set in both the input and the output
data.

A.4.5 Compare received signals with EMS state estimation results and confirm that the phasor system
reports the same values within the following limits:
1. Voltages – 2% magnitude, 1° phase angle.
2. MW & MVAR – 5% magnitude and same direction.

State Estimation (SE) uses primarily watt and var measurements along with bus voltages to estimate the
complex system voltages (magnitude and phase angle) across the grid. By using over-determined
equations and complex processes to reduce errors, the techniques can provide highly accurate results,
including location and reduction of actual measurement errors. SE solutions are used to set operation
guidelines as well as system security assessment, so they become a critical element in system operation.

Page A-10
Phasor system measurements should compare accurately with SE, and differences need to be resolved.
SE is also the only method other than phasor measurement that can provide system phase angles, which
is the basis for power flow in the grid.

 Method

SE is usually performed automatically on a scheduled basis, with the repetition interval ranging from one
to several minutes. It uses measurements gathered primarily with a SCADA system. The comparison for
this test needs to be done using a snapshot of phasor data corresponding to the same time as the SE.
Data for the SE will usually span several seconds to several minutes. The best approach is to use phasor
data with a time stamp close to the middle of the time span used for the SE. It is also important to use a
time when the system is as quiescent as possible. If one can observe the frequency signals, choose a
time when all the phasor frequency signals are near nominal and changing very little. At these times the
phase angles and consequently power flows are undergoing minimum change.

Use a single phasor data snapshot for the comparison, using only valid, synchronized data. Use the
positive sequence voltages and currents for the calculations and comparisons. Use the voltage phase
angle to calculate phase angle between busses. Compare the SE solution at as many points as possible,
making at least one magnitude and angle comparison at every station where there are phasor voltage
measurements. This process is an important step in the validation of phase angle measurement in each
station.

The MW and MVAR comparisons are not as important at this step since they are also covered in A.4.3. If
the SE provides MW and MVAR estimates that are corrected from those supplied by SCADA, compare
these with the phasor provided values as described in section A.4.3. The same observations provided in
A.4.3 about limited accuracy due to low currents apply here as well.

Problems in SE solutions are well documented. The phasor solutions are likely to be better than those
from the SE in boundary areas and where there are limited measurements. This step has also proven to
locate errors in SE configuration. Differences in excess of the given comparison limits should be
investigated to determine the cause, which could be in either measurement system.

A.4.6 Control center installation confirmation documentation

The documentation requirement is the same as for substations confirmation described in section A.3.6.

Page A-11
APPENDIX B: TROUBLE RESOLUTION FOR INSTALLATION AND MAINTENANCE

B.1 Introduction

This appendix provides troubleshooting guidance for both the installation and maintenance phases. The
installation section focuses on the PMU since that is the most unique part of the system installation.
There are often difficulties with communication installation as well, but this is most often with the PMU
communication interface or the networking rather than the communication system itself. Since these
problems are very system specific, they are not addressed here.

Once the PMU is correctly installed and the system is operating, problems are usually detected by the
end user. Often it is difficult to determine where the problem originates since there are many elements
in the system. This appendix provides some guidance for locating and resolving common problems.

B.2 Installation problem solving

B.2.1 PMU installation

The magnitude of voltage and current signals is determined from the sinusoidal AC signal itself; the time
reference does not affect the measurement. The value should be the Root Mean Squared (RMS) value
for the signal. All three phases of a three phase signal and positive sequence should be about the same
magnitude. Phasing can affect positive sequence, so if it is significantly different, check the phasing.

The PMU may require the phases for a specific 3-phase input to attach to specific terminals or may allow
the user to select the phases in software. Errors in phasing can be determined and corrected as follows:

1. Phase order - If A-B-C phases are not in the correct order, the positive sequence will be small, close
to zero. Swapping any two phases will correct the order. A-phase should be the about same phase
angle as positive sequence which can give a guideline for the overall designation.
2. Missing phase - If there is a missing phase, the positive sequence will be 2/3 its normal value. Single
phase phasors should tell which phasor input is missing. Note that a missing input can be from a
connection problem or a failed input channel.
3. Reversed phases – Most PMUs use Y connected voltages so the connection is unlikely to be
reversed. With current, the CTs are line current which can be connected with either polarity. While
there is no absolute designation, the most common orientation is positive power (V x I*) is power
flowing out of the bus. Thus, if using the convention of positive power is power flowing out of the
bus, the current will be approximately in phase with the voltage. Conversely if power is flowing into
the bus, the current angle will be approximately 180 degrees from the voltage. All currents on lines
with the same power flow should have similar phase angles.
4. Designated A-phase – Since measurements cover a wide area, it is necessary to have the same A-
phase designation for all substations over the area where the measurements will be compared.
There is no fixed reference that can be used at a single substation for checking the system phasing

Page B-1
 relative to other stations; it must be checked with other time-synchronized measurements. This is
most easily done at the control center where all signals are available.
5. Y-Delta – phasor measurements are usually referenced to a Y connection for voltage. Currents are
usually phase currents. If some connections are Y and some Delta, there will be 30 degree offsets
and magnitudes to correct. It is best to make all measurements the same. In some cases it is not
possible to access phase currents or Y voltages. In these cases it will be necessary to correct the
angle measurements with a +/- 30 degree correction and the magnitude with the correct sqrt(3)
factor.
6. Incomplete connection – if a 3-phase connection is not accessible, the measurement may be made
with one or two phases. Without all three phases the measurement will not be as good quality, but
this would make a minor difference in most cases. As long as the inputs are connected to their
designated phases, the reported angles will be correct. The amplitude of the sequence quantities
will be 1/3 (one phase connected) or 2/3 (2 phases connected) of the normal value. Correct this
with a PT/CT ratio adjustment.

Frequency is generally derived from rate of change of phase angle. It can, however, be derived in a
more traditional way, such as the period between zero crossings. It can be derived from any AC signal,
but a voltage signal usually one of the individual phases or the positive sequence. The particular source
signal is usually assigned by the PMU vendor. Some PMUs allow the user to select the signal. There may
or may not be a fail over signal designation. If the signal fails, the frequency measurement will be lost.
Some PMUs lock the frequency output to a nominal value if the source is too low; others will continue to
estimate frequency from whatever signal is present, which usually presents a rather poor measurement.
If the frequency measurement appears incorrect or noisy, check that the signal source is properly
selected and active.

Rate of change of frequency (ROCOF) tends to be a very noisy measurement since it is the second
derivative of the phase angle, the usual measured quantity. It is not widely used due to the noise
content. The new standard, C37.118.1 imposes new limits on this measurement and makes some
recommendations for its measurement, so it is expected to improve. It has been observed that when
the PMU is near a device that distorts the waveform such as a Static VAR Compensator (SVC), frequency
and ROCOF are very noisy. If the measurement seems excessively noisy, check the input levels and the
frequency. However the user needs to understand that a ROCOF measurement with high noise may be
inherent with a particular PMU.

The synchrophasor standard, C37.118.2, includes an analog signal type which is not phasors. This is
intended to cover scalar measurements such as exciter values, controller settings, or air pressure. By
sampling these values and including them in the data stream, auxiliary equipment operation can be
analyzed right with the power system. These values can be supplied in many different ways ranging
from sampling by an Analog to digital (A/D) input in the PMU to digital data transfer from other devices.
The measurement characteristics depend on many external factors and therefore are not specified by
the standard. It is left to the user to provide and define scaling, operation, and meaning for these
signals as well as troubleshooting.

Page B-2
The C37.118.2 synchrophasor standard also includes a digital data type which is in Boolean format.
These values normally represent switch, alarm, or other indications that are represented as a 0 or 1.
They are packed into 16-bit digital words for transmission to the control center. In most cases, the PMU
samples the digital input when it sends a report. If the input changes state twice (e.g., 0 to 1 and back to
0) in the same sampling interval, the change will be missed since the value is reported at the sampling
time. In some cases, the PMU may latch the change so it will catch momentary transitions. That leads
to another problem where the reported value is not the current state. These details of performance and
operation need to be examined by the user so that measurement reports can be correctly understood.
If the digital indications seem to be incorrect, troubleshooting consists of checking the input compared
to the output indication. Check that the input voltage matches the threshold for the PMU. In some
cases, the PMU requires an active voltage that exceeds a threshold, perhaps 24 V or 60 V for a 48Vdc or
120Vdc system respectively. In other cases, the input requires contact closure.

Page B-3
B.3 OPERATION PROBLEM SOLVING

These procedures are oriented to solving problems in operational systems. These same methods can be
used for solving problems during system development as needed.

B.3.1 Overview of synchrophasor data using C37.118 transmission

When using C37.118 format, data is sent in packets called frames. Frames are sent at timed intervals
based on a reporting rate that is a sub-multiple of the system frequency. The most common rate used in
North America is 30/s, so a frame is sent every 33.33 ms. Every frame contains a sync word, a time
stamp, a block of data, and a CRC check word at the end. If the frame is from a PMU, the data consists
of a status word, phasors, analogs, digitals, and frequency. If the frame is from a PDC, it may have data
from several PMUs. Data from each PMU is included as a block in the message, each of which has the
aforementioned elements. The timestamp in the frame applies to all the data in the frame, so if data is
combined from several PMUs, it needs to be time aligned. Each PMU block retains its status word,
which validates that particular measurement information. See figure B.1.

Figure B.1. C37.118.2 format data packet showing status indications

The status word contains six basic quality indications. These are data valid, PMU error, sync valid, sort-
by-arrival, data modified, and time quality. The data user has to check these indications to know if the
data item is indeed valid data and if there are restrictions for its use. This quality flag is originally set by
the PMU but may be updated by other devices such as a PDC as the data is streamed.

Page B-4
The data valid bit indicates if the data in the given PMU block is valid or invalid. If the bit is set to 1
indicating the data is invalid, the user should not use the data for anything. At best, the data may be
used but only with a prior knowledge of why the bit is set to invalid. The bit is usually set to invalid by a
PDC to indicate that no data was received from the data source for this particular data frame. In that
case, the data that is in the PMU block consists of random values used as a placeholder. Other reasons
for setting the bit to invalid are a PMU in test mode, receipt of a CRC error, and PMU transmission
problem. Missing or errored data is (supposed to be) set to a NaN (not a number) state to differentiate
non-data from errored data, but this is not done consistently and there is some variability in how NaN is
actually indicated.

The PMU error bit is reserved for the PMU to indicate there is a measurement or operation problem.
The exact meaning of this bit is left to the particular vendor since there are many problems it could be
used to indicate. These can include A/D problem, computation overflow, memory failure, input failure,
configuration error, etc., but are not limited to these issues. The PMU error may be fatal and invalidate
some measurements or may be an advisory. In any case, when this error is detected, the user should
investigate and determine the cause of the indication before using the data. Note that some vendors
have used this flag for other things relating to data processing as well. The user needs to follow the data
chain to find the source of the flag and take action accordingly.

The sync valid bit indicates whether the measurement is accurately synchronized to UTC time. It is set
to 1 when the time is not synchronized to UTC. The time source must provide an indication to the PMU
when it is not in sync. If the time source is a GPS receiver, the PMU can read a message indicating if the
receiver is locked to UTC. If the receiver is external and supplies time to the PMU using a time code such
as IRIG-B, it needs to include an indication which could be part of the code or supplied by another signal.
Once the bit is set to 1 (not in sync), it should never be set to 0 by other processing in the chain since no
device can re-synchronize it. If the PMU reports that the measurement is in sync (bit cleared to 0),
succeeding devices will normally pass on the flag without changes. However, it is possible that the PMU
incorrectly indicates that the measurement is in sync; if a succeeding PDC determines the measurement
is not in sync, it may set the bit to indicate not in sync. Note that if the measurement is not in sync, the
phase angle will not be dependable, but the magnitude measurements should be intact. Generally, the
frequency and ROCOF will be within usable limits as well. Also note that in most cases, when all
measurements from a single PMU use the same internal time reference, the angles between phasors
from the same PMU are accurate even though the external sync is bad and the angles with phasors from
other PMUs are invalid.

The sort-by-arrival bit indicates that the data has been assigned a local timetag by the receiving device.
In the case that a PDC receives data with a time stamp that is not reasonably close to the current time,
the PDC can detect this as a timetag failure and assign a time stamp locally. In this case it must set the
sort-by-arrival bit to indicate the timetag is artificial so it will not be misinterpreted by subsequent
systems. It is also recommended that the sync valid bit should be set to indicate the synchronization is
invalid, since the phase angles will not be reliable with any timetag change.

Page B-5
The rationale and process for sort-by-arrival is as follows: Phasor data systems send data in real-time
with minimal delays. Generally the time delay from the measurement until it is received by the first PDC
is a fraction of a second. The timetag applied to the data by the PMU is derived from a clock which
should be very precise, but could be in error, even a very large error. If data is received by a PDC has a
timetag that is clearly incorrect, the time that the data is received at the PDC is closer to the actual
measurement time than the timetag supplied in the data frame. In this case, PDC assigns a timetag
based on the time that it receives the data. This new timetag, based on when the data arrives, will likely
be within a fraction of a second of the actual measurement time. This process was executed by placing
the data in a frame with other data corresponding to the most current time (i.e., “sorted into the most
current data based on when it arrived”).

Sort-by-arrival (SBA) or local timetag processing is used when there is no synchronizing clock or when
the PMU has a known bad clock. It is also occasionally used when there are long delays, such as playing
back old data for test purposes. The process can be used at higher level communication, such as PDC to
PDC communication, not just directly from a PMU. Some PDCs will automatically revert to SBA from
timetag alignment when the time offset exceeds a certain value or when the sync bit is set, and may also
recover to time alignment when the problem clears. These parameters may be user settable. Having a
SBA or local timetag option allows maximum data utilization even when the measurement system is
impaired. The data modify bit indicates that the data has been modified by another device such as a
PDC. This should be set whenever any data has been modified such as the magnitude rescaled or the
angle rotated. It should also be used if the data was inserted by and interpolation program.
Unfortunately with a single bit for the whole data packet it is impossible to clearly indicate exactly what
was modified. This bit simply warns the user that they are not seeing data as originally sent and if there
is a concern, they will have to investigate.

The time quality bits provide the accuracy of the timing source used by the PMU at the time the
measurement was made. Originally only the sync bit was available to determine that the timing was
locked to the reference within a usable range. The range was not specified, but most vendors used a 1-5
microsecond range to determine the time was locked. This 3-bit time quality indicates the actual time
quality range since range of locked can vary. The 7 ranges indicate 0.1 to 10,000 microseconds, a good
match to phasor measurement.

B.3.2 Operational problem troubleshooting

This section discusses a variety of aspects that may be observed in the data that are the result of
installation, setting, and communication decisions as well as the result of errors or failures in the
measurement system. Each section describes the cause, observable effect, and a variety of changes that
will remedy a deficient situation.
B.3.2.1 Phase Loss

Normally each input will be three phases in approximate balance. Loss of a single phase will result in a
large imbalance. The amplitude of positive sequence will drop by 1/3, negative and zero sequences will

Page B-6
increase by the same amount, and the lost phase will go to 0. Note that the positive sequence phase
angle will not change noticeably. Single phase reporting is the best way to detect this kind of problem,
but reading 2/3 of normal in positive sequence is also a good indication. Unless there are
complimentary measurements nearby, it is not possible to determine from measurement alone if the
problem is primary system failure or something in the measurement chain. However, this primary
system failure would likely be detected quickly by protection equipment, so usually this type of failure
can be analysed as a measurement system problem and resolved by on-site investigation.
B.3.2.2 Resolution & Accuracy

Two of the primary aspects in a measurement system are accuracy and resolution. Accuracy refers to
how closely the measurement compares to an accepted, standardized value. That is, if the values
indicated by a measurement system will report the same value that a calibrated measurement
instrument would report. Resolution refers to how finely a measurement is made. In digital terms, it is
the number of bits of resolution. It should be noted that a measurement can have high resolution but
poor accuracy. Conversely it can have high accuracy but poor resolution so it is difficult to determine if
it is accurate.
For example, a VOM with 1% accuracy may report a reading of 1.20564 V. In this case, 1% of the reading
is .012, so the .00064 is not usable for determining the actual voltage. However, if it can make
consistent readings at that level (that is, if reading the same voltage several times will result in the same
number), it is clear that reading a different voltage and with a measurement of 1.20521 is a lower
voltage. Here the resolution is good, but comparison with a calibrated standard is limited to 1%.
For a converse example, a meter with 0.1% accuracy may report a measurement of 1.2 V. In this case,
the accuracy is good to .0012 but since the meter only reports to .1, all digits should be accurate and
true, but the meter cannot be used to its full capability.
Phasors are computed from sample values. Sample values have a fixed resolution, usually 16-bits, but
since phasors are computed they can have a much higher resolution. If transmitted as integer values,
the resolution is limited to 16-bits (C37.118 protocol). Here the scaling will make a big difference in the
apparent resolution. If the transmitted value is nearly full scale, then the value will have full 16-bit
resolution. If the value is near 0, then there may be just a few bits of resolution.
As an example, if the phasor represents a 230 kV voltage, the transmitted value will be the phase to
neutral voltage which is about 132,800 V. The maximum 16-bit signed number is 32767. If the value is
scaled to read 132,800 when the integer value is 32767, each bit of the integer value represents 4.05 V.
If the voltage went up to 132900, the integer value would overflow and it could not be transmitted.
Consequently a scale factor is required so that a voltage as high as could be reasonably encountered can
be represented by the integer value, but no larger so that the resolution is as good as possible. If a
scaling of 8 V/bit was applied, then we could represent 32767 x 8 = 262 136 Vp-n (454 kV p-p for the 230
kV voltage) which should be adequate. The resolution is 8V for a nominal 132,800 V signal which is
.006% of the reading. Put in a different context, if viewing a plot of one minute where the steady-state
voltage varies 500 V, the trace would look like a nearly smooth line.
Contrast that with a scale factor of 1000 V/bit. Then 132,800 V would be represented by a phasor value
of 1328 which has a resolution of .0075 or .75%. A phasor value as high as 3,276,700 V could be
reported which is much higher than necessary. If viewing a plot where the steady-state voltage varied
500 V, the trace would look like a straight line unless the value crossed a bit threshold where there
would be a 1 bit step. It would be difficult to make meaningful comparisons of voltages scaled this way

Page B-7
in any except large voltage swings. Low resolution also limits the ability to do small signal analysis since
the signals are lost in truncation.
Most PMUs internally compute phasor, frequency, and ROCOF values using 32-bit integer or floating
point math. This gives a high resolution for internal computation and representation. Transmitting
these values using floating point format will maximize the available resolution. This will provide the
greatest ability to compare signals and derive low level information from them. Transmission in integer
format limits the resolution to 16 bits and depending on the scaling, can considerably reduce the
derived resolution. However it probably will not have much effect on accuracy since accuracy is
generally 0.1% to 1 % which can be represented with only 10 bits of resolution. The down side of using
floating point representation is that the numbers are twice as large requiring about double the
bandwidth and data storage. Properly scaled integer values can provide the same accuracy in most
cases with minimal loss of resolution. The pros and cons need consideration when choosing the data
type for data reporting.

Figure B.2 Frequency reported in integer values

This plot shows frequency reported in integer values. The C37.118 standard requires frequency
reporting in mHz, so the resolution is .001 Hz. In this plot where the frequency is very steady, the
discrete .001 Hz steps are clearly evident. The frequency is probably computed to a much higher
resolution but rounded off for reporting. Since .001 Hz is a very small part of the nominal 60 Hz value,
this loss of resolution is likely to be insignificant, but in this case it looks rather coarse. Frequency
measurement accuracies are typically .001 to .004 Hz.

Page B-8
B.3.2.3 Input Scaling

Voltage and current inputs need to be scaled so that the input is not saturated with the highest
amplitude of interest. Conversely, it should be set so that there is as much resolution as possible for low
input values. In the case of voltage, it probably should be set so that the maximum input is 1.5x to 2x
the nominal value. This should cover voltage swings and still give good resolution at nominal and
depressed voltages. Current is more difficult to set. For most phasor monitoring uses, setting the
current input for the current that will be seen with 2x the rated power for the line will capture most
swings and give good resolution at light load (0.1 of nominal), though it is probably better to set it a little
higher to be sure to capture swings. However in some cases, more resolution may be needed at low
levels and in other cases more range for overloads may be required. Scaling should be adjusted
according to the typical operation of the line.
The consequence of input scaling errors is noisy or inaccurate readings. If the input exceeds the A/D
input range a small amount, the phasor should be derivable with a good phase angle and acceptable
magnitude even though the AC waveform will be clipped. If the waveform significantly exceeds the
input capability the phasor magnitude will be reduced and at some point the angle will be inaccurate. If
the input is very low compared with the A/D range, there will be few bits of resolution and the
measurement will become noisy and inaccurate. Handling over-range and under-range conditions and is
highly dependent on the particular phasor algorithm.
B.3.2.4 Output Scaling

The PMU calculates phasor values from input waveforms. Internally it uses floating point or integers of
32-bit or greater resolution, so the phasor values are generally derived with high resolution. When the
PMU outputs data, it converts the internal numbers to either IEEE 32-bit floating point or 16-bit integer
values (±32,767). With floating point, the resolution should be as good as the internal derivation and
certainly much better than the accuracy limit. With integer values, the value appears as an integer, so a
scale factor must be applied to keep the value within the integer range. The nominal C37.118 scale
factor is an (integer) × 10 . This scale factor should be chosen so that at the maximum phasor value
times the scale factor is the maximum value that should be reported. This gives the maximum
resolution for lower values and the widest range of reporting. For example, we may want to allow for a
100% overload for a 230 kV voltage. That is, we want to be able to see a voltage up to 460 kV. Voltages
are transmitted as phase to neutral voltage, so here we want to report up to 265,600V (nominal
132,800V p-n). This gives the scale factor PHUNIT = (265,600/32767) × 10 = 810571. Each phasor
value will be multiplied by 8.10571 to display the line value. The resolution is thus 8.1 V/bit (14 V/bit l-l),
which is very good for a 230 kV system. A larger scale factor will give more maximum range but less
resolution for small values.
As a general recommendation, scaling for synchrophasor reporting should be set to report systems
voltages up to 1.5 to 2 times the nominal line voltage. This allows observing the overloads and swings
that may be encountered in almost all situations and still gives good resolution of the signal. While
higher voltages may occur during some extreme events, these are rare and usually only of an interest in
protection type analysis. Scaling for currents is normally best served with a maximum value 2x-4x the
rated load. Most lines carry considerably less than rated capacity most of the time, and often have light
loading. At 2x rated load, the resolution at 10% load is still 1/1600 or about .05%. In some cases scaling
at greater than 2x rated load is needed, so the scaling needs to be adjusted accordingly.
Not all PMUs allow the user to scale the outputs to their own specifications. In these cases the scaling
should be reasonably set for resolution and maximum range, but the user should check to be sure the

Page B-9
settings meet their needs. Analog data, as defined in C37.118, is used for data types other than phasors,
such as MW and MVAR measurements. These data types are not specifically identified and
consequently their scaling is not discussed here. Frequency data in C37.118 is always scaled in mHz
(0.001 Hz) and rate of change of frequency (ROCOF) is 0.01 Hz/s. In all cases, using floating point
representation provides the best solution for limitations in maximum range versus resolution.
B.3.2.4 Sync Loss at a PMU

The phase angle measurement is determined in reference to time. Angles can be used together only if
they are made with the same time reference. There are two important considerations here. First, all
measurements that use a common time reference have phase angles that can be compared with each
other. For example, if several measurements are made by a single PMU these should be synchronized
to each other so the phase angles are comparable with each other. Likewise a group of PMUs served by
a single clock will have comparable angles even if that clock loses UTC sync. This is useful in computing
power from one or more PMUs in a substation. Second, the time reference (GPS clock or PMU with a
GPS receiver) will have a local clock “flywheel” which will dictate how the time-angle reference will vary.
A highly stable clock can remain synchronized to UTC within a few microseconds (µs) for hours or even
days, while an undisciplined crystal oscillator can be off by several milliseconds in just a few minutes. At
60 Hz, one degree of phase angle is about 46 µs. A 5 µs error is only about 0.1 degree of error which is
generally very acceptable. At 100 µs of error, the phase angle measurement is not very usable. The
C37.118 standard requires the PMU to set a bit indicating external UTC synchronization has been lost
within 1-minute of the actual loss. This warns the users that the angle is not a reliable measurement as
long as that bit is set. However with C37.118.2 there is also a time quality measurement. If this feature
is properly implemented, it could be used to determine if the time accuracy is good enough to continue
marking the data as synchronized rather than simply concluding that lock is lost. Thus the PMU can use
this for controlling the sync bit.
A measurement with sync loss may not be evident quickly, since it takes some time for the phase angle
to drift away from an accurate value. Typically the time drift is one direction, positive or negative, and
appears as a drifting phase angle. Since system noise and normal angles are generally larger than the
drift, it may be hard to determine quickly, but will be observable over time. In the case that the sync
loss is not detected or properly flagged, the user needs to be on the lookout for phase angles that have
changed or do not match expected values.
B.3.2.5 Intermittent Sync

Intermittent sync is sometimes more difficult to detect than a complete loss. Occasional short loss of
lock to GPS is a normal occurrence and should be transparent to the user. These can be the result of
switching synchronization between satellites, a short burst of noise, antenna shading, or other effects.
These should be short—up to a few seconds—and random. They are generally short enough that they
are not even reported by the clock or PMU. However, outages may occur that are significant enough
that the clock cannot carry through the outages. Intermittent sync can also be caused by bad
connections or signal mismatch. In either case, the sync loss bit should indicate there is a problem. The
signal measurements may only show very small drifts followed by a drift back to accuracy or can show a
snap back to correction when the clock re-syncs. The particular behaviour depends on the clock
implementation. The main points are to monitor the sync loss bit to detect developing problems and
keep in mind the effect of synch loss when troubleshooting.

Page B-10
B.3.2.6 Derived Signals

Frequency is usually calculated as the derivative of the phase angle measurement. ROCOF is the
derivative of the frequency. The derivative emphasizes changes, so if the base measurement is noisy
the derived signal will be very noisy. However, these signals are also good indicators of change. A fault
may cause a sudden change in phase angle which in turn causes a jump in frequency. Any sudden
change in frequency will cause a spike in ROCOF. Because ROCOF is a double derivative, it tends to be
too noisy for useful analysis, but it serves as a good indicator of a disturbance. The new standard,
C37.118.1, sets requirements for both Frequency and ROCOF, so the newer PMUs should have more
usable and steady reading for these measurements.
B.3.2.7 Delays in signals received at the Control Center

Delays are usually small, but can be up to several seconds. This is highly dependent on the
communication system. Since data is sent on a continuous and timed basis, it is possible to confirm the
output delay from the PMU itself. If the PMU meets C37.118.1 requirements, it is required to output
data within a few reporting periods (see standard). Using that as a baseline, a receiving device can be
used to timetag when the data is received at the control center, thereby testing the communication
delays. While an accurate measurement is not necessarily easy to perform, an approximate value that
compares delay with other data inputs is quite achievable. The important part of this procedure is to
make sure that delays from all remote units are similar. It is best to do this test in over a 24-hour (or
longer) interval to be sure that there are not some periodic events that adversely affect one or more
inputs. If this or other continuous monitoring is not feasible, 1-5 minute snapshots throughout a 24-
hour period are advisable.

Differences in delay between signals should be less than 100 ms. unless very special circumstances
apply, such as separate back-to-back connected communication systems or very big distance variations.
Differences over the 24-hour period should show smaller changes, perhaps 50 ms. or less. In either
case, results that are significantly different than these general time durations should be investigated.
There could be perfectly acceptable reasons for different numbers, but generally larger deviations are
because of a problem in the configuration.

B.3.2.8 Data loss in signals received at the Control Center

Data loss is the most common problem in data transmission from the PMU to the control center. The
cause can range from faulty equipment to overloaded communication channels. Some data loss is
acceptable as there always are random error effects. Generally the data loss should be less than 0.1% in
all cases. Loss < 0.001% is very achievable.

When significant ongoing data loss occurs, there may be a pattern that will help determine the cause.
Patterns could include a single sample at regular intervals, blocks of samples that are the same size but
randomly spaced, or something else periodic. Random single samples at rates < 0.01% (1 sample/10,000
equivalent to 1 sample/6 min at 30/sec) is not worth pursuing. 0.1% (2 samples/min at 30/sec) may be
acceptable in overall performance, but it shows communication problems if it occurs continuously. At
this rate of loss, overloaded network segments or faulty equipment are likely to be the problem. If it

Page B-11
happens on a regular interval, look for a correlation between the time and the data loss. Greater
amounts of data loss are likely due to communication overload or even network mismatches (such as
half vs. full duplex). It is often easier to diagnose problems with larger amounts of data loss. In all cases,
look for regular patterns of loss and correlation between time and operation of other systems. The
problem could be at (or in) the PMU, so it may be necessary to intercept and check the data loss at the
PMU output as well as at points in the data transmission system.

B.3.2.9 Data errors in signals received at the Control Center

Data errors are here defined as receiving packets of data as expected, but with an error such as
incorrect packet size, incomplete packet, packet run-on with another packet, and packet with CRC error.
Errors in data transmission are usually masked by the receiving systems as data loss. Most phasor
systems now use a network communication system which protects data integrity at the data link layer
with a CRC or other means. When that layer receives data with an error, it discards the packet and the
user only observes a missing packet.

If a packet is received that is the wrong size but otherwise complete, it indicates the sending PMU has
changed configuration but the PDC has not detected it and reconfigured. This is a mismatch and is easily
resolvable, the method depending on the PDC technology. If the packet is incomplete or is run-on with
other packets, it indicates the system is fragmenting the data and it is not being reconstructed correctly.
This will usually only be seen where the packets are large and an intermediate link needs to fragment
packets. Then if pieces of fragments are lost, the whole message will be unusable. Consider improving
the intermediate links or reconfiguring to send smaller packets. When using network communications,
packet CRC errors will be rarely observed because of the data link layer CRC.

Systems that still use serial communication have direct access to the RS232 interface and detect errors
through the packet CRC. These can be observed by the users. Generally such errors are only caused by
data collisions or interrupted channels. If data errors of this nature happen more than a few times/day,
the complete communication channel needs end to end testing to find the problem. Alternatively there
could be a modem problem. It is possible that there is a problem in the PMU or receiving device itself,
requiring more thorough testing to determine.

Page B-12